Quantcast

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Advice on 'monster' set-up

2

Comments

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    If it's purely frame rates per dollar, there are things you could do to get, say, 2/3 of the performance for half of the price.

    One tweak that I would make if you're buying from New Egg is to get the Mushkin Callisto Deluxe SSD instead of the Corsair Force one:

    http://www.newegg.com/Product/Product.aspx?Item=N82E16820226152

    Same SSD controller, same NAND flash, etc., so there's no chance that the Mushkin SSD will be worse than the Corsair one.  Corsair has done rather questionable things with their firmware in the past, so I'm not sure if the Mushkin and Corsair SSDs would perform exactly identically, or if the Mushkin one would be better.  Either way, it's $28 cheaper for the same capacity, which makes it a better deal.

    Don't forget that you need an OS license if you don't already have it.

    http://www.newegg.com/Product/Product.aspx?Item=N82E16832116986

    There are also professional and ultimate editions of Windows 7.  Professional makes sense for businesses, as some of the features in it are critical for keeping dozens of computers in a business working together, but is a waste of money for the overwhelming majority of home users.  Ultimate is a waste of money for the overwhelming majority of everyone, and I think the main reason why it exists is so that people who don't know any better will have a way to give Microsoft more money for no good reason.

  • RobgmurRobgmur Member Posts: 322

    Microsoft Windows 7 Home Premium SP1 64-bit  , I think i can get it for Free, not sure yet. No i see no point in paying hundreds for an OS

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • paroxysmparoxysm Member Posts: 437

    If you want some more thought on comparing models and performance, there are a lot of hardware sites out there that test and compare hardware in many ways.    Best at a certain price point. complete comparison of certain hardware in control systems, etc.  I "personally" like tomshardware for the sheer amount of tests and comparisons they do.  There are a lot of other hardware sites you can visit as well and I'm sure people will list them for you as well.  More knowledge is never a bad thing.

     

    You can also do a decent comparison of models on some manufacture sites.  Like you, I am currently looking at a new Sandy Bridge system.  I've been doing nothing but Asus boards for a long time.  I used to do Tyan server boards before that, but that was mostly for mulit proc systems.  If you go to the Asus MF site, you can use the built in compare tool to find which of the plethora of boards have all the options you want.  It even shows which boards allow which speeds(x16/x8/x4/x1) when put in crossfire/sli setups from 2-4 cards.  Write down the products you are considering and then look for reviews and comparisons of those models.  It's a lot to consider, but I like to try to make well informed decisions.

  • RobgmurRobgmur Member Posts: 322

    Sounds good, thank you.

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • paroxysmparoxysm Member Posts: 437

    Not really related to your setup, as you are aiming for higher level cards, but I just read that some AMD new cards are just rebadged. 

    6770 = rebadged 5770

    6750 = rebadged 5750

     

    http://www.anandtech.com/show/4296/amds-radeon-hd-6770-radeon-hd-6750-the-retail-radeon-5700-rebadge

  • RobgmurRobgmur Member Posts: 322

    Say I get trifire 1x 6990 and 1x 6970, that would cut the noise of the 6990?

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • psyclumpsyclum Member Posts: 792

    if you are actually looking for THE monster system, may i suggest you look into

    http://www.newegg.com/Product/Product.aspx?Item=N82E16820227706

    assuming you will be using windows7 (sandforce works better with win7)

    here is the review for it.

    http://www.anandtech.com/show/4256/the-ocz-vertex-3-review-120gb

    here is a 2nd opinion

    http://www.tomshardware.com/reviews/vertex-3-sandforce-ssd,2869.html

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    Originally posted by Robgmur

    Say I get trifire 1x 6990 and 1x 6970, that would cut the noise of the 6990?

    No.  Then you'd just have the noise of the 6990 plus the noise of the 6970, which would be more noise than the 6990 alone.

    You could do three Radeon HD 6970s in CrossFire, if you want more than two.  I think that's pointless on a single monitor, though.

    -----

    "If you want some more thought on comparing models and performance, there are a lot of hardware sites out there that test and compare hardware in many ways. Best at a certain price point. complete comparison of certain hardware in control systems, etc. I "personally" like tomshardware for the sheer amount of tests and comparisons they do. There are a lot of other hardware sites you can visit as well and I'm sure people will list them for you as well. More knowledge is never a bad thing."

    Different sites are useful for different things.  Tom's Hardware's best video cards for the money lists have some merit, though they're way too fond of CrossFire and SLI.  Their best processors for the money list has considerably less merit, as they ignore motherboard costs, which are a huge factor.  Their best SSD for the money list is a complete joke.

    Different sites are useful for different purposes.  For processor reviews, any major site will do, but the reader needs to be able to distinguish between single-threaded and well-threaded benchmarks to understand what you're reading.  They're not necessarily labeled as such, but you can deduce it from the results.

    For video cards, a lot of sites will run canned benchmarks and give you about the same information.  Hard OCP is the indispensible video card review site, as they'll actually play real games, and see which cards let you turn video settings the highest.  If in a particular game at particular settings, card A offers 250 frames per second, which card B only offers 200, that's not useful information, as they're both capped by your monitor refresh rate.  If card A offers offers 50 frames per second and card B 40, that's more useful, as that difference matters.  The first situation does not imply the second.  Just avoid X-bit labs, as they benchmark AMD cards using SSAA rather than MSAA for some inexplicable reason.

    For solid state drives, AnandTech is by far the best.  Legit Reviews has some useful reviews for SSDs, too.  Some sites will run a bunch of sequential benchmarks and can't figure out why JMicron SSDs have a stuttering problem, which is completely useless.  Since you mention Tom's Hardware above, I'll add that that particular site hasn't figured out what to do with SSDs yet.

    It's hard to find good motherboard reviews, simply because no site has the time to review more than a tiny fraction of the motherboards on the market.  You can amalgamate a lot of reviews to figure out that Asus and Gigabyte are the best brands, with MSI and AsRock the others worth considering.  If you get a motherboard from ECS, Jetway, Foxconn, Intel, or Biostar, then there's a pretty good chance that there will be something serious wrong with it.  EVGA used to be a good motherboard brand, but then their motherboard team got hired away by Sapphire, which doesn't yet have much of a track record.

    For power supplies, try Jonny Guru, Hard OCP, or Hardware Secrets.  Some sites try to review power supplies but don't have the equipment to test them properly, which is useless.  I'll mention that if Hard OCP rates a power supply as "fail", and Hardware Secrets gives it a "silver award", those aren't contradictory at all.  It's a difference in editorial standards.  Hardware Secrets gives their golden award an awful lot, but you can still get good information from their testing results.

  • ShinamiShinami Member UncommonPosts: 825

    I just tested a Caviar Black series against a Raptor/Velociraptor series.

     

    Its true that a Caviar Black drive is cheap, but if you depend on playing Roleplaying Games which have cinematic scenes or file loading while while within a map like in most MMORPGs...you will find stuttering a slowdown. 

     

    Due to how some people in this forum can not shut up about Caviar Black hard drives, I tested myself several Caviar Black drives in both of my machines (My crossfire and SLI system, both under Windows 7 64 bit). I tested them against 150GB, 300GB and 600GB Raptor/Velociraptors...

     

    What I found was that in Cinematic scenes between characters where you see a dialogue exchange taking place between characters in games like Dragon Age Origins or Dragon Age II, you literally see stuttering and some slowdown. Things I never saw when I played on a 150GB Raptor. 

     

    I had near-empty caviar blacks against near full Raptor/Velociraptors.

     

    The 10K RPM as usual DOES make a difference along with the reliability. If you are creating an ELITE build which is a MONSTER PC, Do not settle on 7200 RPM HDDs unless they are meant for Archiving DATA. Raptor/Velociraptor are ENTERPRISE Class Drives capable of running 24/7 without them blowing out on you. 

     

    .and yes...I've tested SATA II and SATA III versions of Caviar Blacks and they have the same problems. 

     

    Now these tests were for Singleplayer Roleplaying Games and MMORPGs. 

     

    I found in testing Real Time Strategy and First Person Shooter games that while the drives could load their files pretty fast into memory, all files tend to be loaded prior to a map starting in an RTS or FPS multiplayer map. In other words, the files are loaded and you wont find yourself loading anything midgame. In short there is almost no real difference in performance between a RAPTOR/VELOCIRAPTOR and a CAVIAR BLACK when playing First Person Shooter Multiplayers and Real Time Strategy Games....(specially in Skirmish and Multiplayer Maps)..

     

    The difference in performance truly is seen anytime a game must load data mid-game (like while in a map.) 

     

    My advice to you is either run a 10K RPM HDD for MMORPGS/RPGs or drop them in your SSD or Ramdisk heavier files.

     

    My testing was ran in both of my systems....and they both displayed the same problems. ^_^ 

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    Well of course a WD Caviar Black isn't as fast as a VelociRaptor.  It's not as fast as a Seagate Cheetah, either.  But it is faster than a Seagate Barracuda, a Samsung Spinpoint F3 or F4, or a Hitachi DeskStar, and that's the real competition.

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    Originally posted by psyclum

    if you are actually looking for THE monster system, may i suggest you look into

    http://www.newegg.com/Product/Product.aspx?Item=N82E16820227706

    assuming you will be using windows7 (sandforce works better with win7)

    here is the review for it.

    http://www.anandtech.com/show/4256/the-ocz-vertex-3-review-120gb

    here is a 2nd opinion

    http://www.tomshardware.com/reviews/vertex-3-sandforce-ssd,2869.html

    While the OCZ Vertex 3 is faster than the Mushkin Callisto Deluxe, both at 120 GB, it's not an enormous difference.  The price gap is an enormous difference, however.  While faster is, in principle, always better, there's a point where you're not having to meaningfully wait on storage anymore, and at that point, it's fast enough.  The Mushkin Callisto Deluxe is already past that point for ordinary home use.

    The difference between 100 IOPS in a given test for a hard drive and 10000 IOPS in the same test for a good SSD matters tremendously.  The difference between 10000 IOPS and 20000 IOPS for two different SSDs doesn't matter much.

    By analogy, if you could reduce your ping time from 100 ms to 5 ms, that would be a huge deal, and well worth paying $200 in a high end gaming system.  If you could then further reduce it from 5 ms to 4 ms, that wouldn't be so important.  Sure, you'd rather have it than not, but that's not $100 worth of nice.

    Furthermore, even if the newer and older SandForce drives were the same price, right now, I'd likely still go with the older.  The Mushkin Callisto Deluxe is based on a first generation SandForce controller, which has been out for a year.  The firmware is demonstrably mature, and the drives will just work right.   The OCZ Vertex 3 is a new SSD based on a second generation SandForce controller.  The firmware will probably work just fine, but it hasn't been in the wild for long enough to know for certain.  If it doesn't work, you could end up with data corruption, a bricked drive, or horrible performance for weeks while waiting for a fix.

    The OCZ Vertex 3, like the other new SSDs to launch this year, is still bleeding edge technology to some degree.  Perhaps to a larger degree than the other new SSDs, for that matter, as it's a new SSD controller, not just new firmware on what is essentially an old controller.  Now, I wouldn't be against a new video card, as you can live with troublesome video drivers for a few months until they get fixed.  But bad SSD firmware is an entirely different critter, as it could make the computer entirely non-functional.

  • paroxysmparoxysm Member Posts: 437

    Originally posted by Quizzical

    Originally posted by Robgmur

    Say I get trifire 1x 6990 and 1x 6970, that would cut the noise of the 6990?

    No.  Then you'd just have the noise of the 6990 plus the noise of the 6970, which would be more noise than the 6990 alone.

    You could do three Radeon HD 6970s in CrossFire, if you want more than two.  I think that's pointless on a single monitor, though.

    -----

    "If you want some more thought on comparing models and performance, there are a lot of hardware sites out there that test and compare hardware in many ways. Best at a certain price point. complete comparison of certain hardware in control systems, etc. I "personally" like tomshardware for the sheer amount of tests and comparisons they do. There are a lot of other hardware sites you can visit as well and I'm sure people will list them for you as well. More knowledge is never a bad thing."

    Different sites are useful for different things.  Tom's Hardware's best video cards for the money lists have some merit, though they're way too fond of CrossFire and SLI.  Their best processors for the money list has considerably less merit, as they ignore motherboard costs, which are a huge factor.  Their best SSD for the money list is a complete joke.

    Different sites are useful for different purposes.  For processor reviews, any major site will do, but the reader needs to be able to distinguish between single-threaded and well-threaded benchmarks to understand what you're reading.  They're not necessarily labeled as such, but you can deduce it from the results.

    For video cards, a lot of sites will run canned benchmarks and give you about the same information.  Hard OCP is the indispensible video card review site, as they'll actually play real games, and see which cards let you turn video settings the highest.  If in a particular game at particular settings, card A offers 250 frames per second, which card B only offers 200, that's not useful information, as they're both capped by your monitor refresh rate.  If card A offers offers 50 frames per second and card B 40, that's more useful, as that difference matters.  The first situation does not imply the second.  Just avoid X-bit labs, as they benchmark AMD cards using SSAA rather than MSAA for some inexplicable reason.

    For solid state drives, AnandTech is by far the best.  Legit Reviews has some useful reviews for SSDs, too.  Some sites will run a bunch of sequential benchmarks and can't figure out why JMicron SSDs have a stuttering problem, which is completely useless.  Since you mention Tom's Hardware above, I'll add that that particular site hasn't figured out what to do with SSDs yet.

    It's hard to find good motherboard reviews, simply because no site has the time to review more than a tiny fraction of the motherboards on the market.  You can amalgamate a lot of reviews to figure out that Asus and Gigabyte are the best brands, with MSI and AsRock the others worth considering.  If you get a motherboard from ECS, Jetway, Foxconn, Intel, or Biostar, then there's a pretty good chance that there will be something serious wrong with it.  EVGA used to be a good motherboard brand, but then their motherboard team got hired away by Sapphire, which doesn't yet have much of a track record.

    For power supplies, try Jonny Guru, Hard OCP, or Hardware Secrets.  Some sites try to review power supplies but don't have the equipment to test them properly, which is useless.  I'll mention that if Hard OCP rates a power supply as "fail", and Hardware Secrets gives it a "silver award", those aren't contradictory at all.  It's a difference in editorial standards.  Hardware Secrets gives their golden award an awful lot, but you can still get good information from their testing results.

    That's why I left it the way I did.  I knew someone would post some other sites.  I myself read Toms, Anandtech, ArsTechnica, and HardOCP.  I was even hoping someone would list a site I didn't know about.  I never base something off just one review and you are right about Toms not covering SSDs well yet, but I think they plan to improve on that.  Also, thanks for your input on the subject and the sites in question.  I do not consider myself an expert on hardware and was just trying to give him options.

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    It's been more than two years since Anandtech demonstrated the right way to test SSDs.  Some other sites have figured out what to do with them in the meantime, but Tom's Hardware isn't one of them.

    If you want to find the best SSD for the money, then the way to do it is to pick your desired capacity, and find the cheapest SSD with a SandForce or Marvell controller of that capacity.  Then you check to see if there's an Intel or Samsung 470 for substantially cheaper, and if so, then get that instead.  Though with Intel, it's awkward, because those hit different capacity points.

    The power supply reviews on Tom's Hardware are comically awful.  A while back, they had a roundup of some high end power supplies.  They didn't even test voltage regulation at all, and that's one of the most important properties of a power supply.  And that was the roundup where they said, okay, okay, we're finally going to test ripple, which we've previously been completely ignoring even though all of the reputable power supply review sites will tell you it's very important.  In their conclusion, they based their recommendations on what to buy largely on the hold-up time, and rejected three of the four best in their review on that basis.  Most sites don't bother to test the hold up time, because if it matters to you, then you should get a UPS, anyway. 

  • RobgmurRobgmur Member Posts: 322

    ASUS MAXIMUS IV EXTREME (REV 3.0) LGA 1155 Intel P67.’ 257$. Better bet for motherboard?


     


    And..


     


    Would it be wise to replace the internal hard drive I had, with this; Western Digital VelociRaptor WD3000HLFS 300GB 10000 RPM as he was suggesting? or add this SSD for gaming ? (-OCZ Vertex 3 VTX3-25SAT3-120G ` 300$) instead of the Corsair Force CSSD-F120GB2-BRKT 2.5" 120GB  ~225$

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • CatamountCatamount Member Posts: 773

    Originally posted by Robgmur


    ASUS MAXIMUS IV EXTREME (REV 3.0) LGA 1155 Intel P67.’ 257$. Better bet for motherboard?


     


    And..


     


    Would it be wise to replace the internal hard drive I had, with this; Western Digital VelociRaptor WD3000HLFS 300GB 10000 RPM as he was suggesting? or add this SSD for gaming ? (-OCZ Vertex 3 VTX3-25SAT3-120G ` 300$) instead of the Corsair Force CSSD-F120GB2-BRKT 2.5" 120GB  ~225$

    A WD VelociRaptor doesn't really make much sense. Either they offer you a little more speed over a traditional hard drive, which is just woefully slow to begin with, for an enormously increased price, or they offer you a bit better price per GB over SSDs for an enormous drop in speed.

    Note, of course, that there's a huge price difference between the newest SSDs and those that have been out for a bit. The Vertex 2 is $199 for 120GB, while the Vertex 3 is $299. Correct me if I'm wrong, but didn't Quizzical have an article just a bit back showing that the Vertex 3 drives weren't really that much faster in real world performance at smaller sizes anyways?

  • ShinamiShinami Member UncommonPosts: 825

    Velociraptors are very strong.

     

    Before one of my Raptors failed on me, I had it run for three years 24/7 under moderate to heavy loads to servers I was running for MMORPGs and other projects. The longest a 7200 RPM HDD ever lasted for a server with a lot of failures was around six months. A 150GB SATA I Raptor from four to five years ago actually runs faster than any current generation Caviar HDD in about every major game. I should know this...I tested this throughly in the last week...

     

    Remember my exception. It runs faster in games where data is loaded after a player spawns (MMORPGs, RPGs) but runs about the same in games where data is loaded prior to spawning (FPS and RTS multiplayer). The issue comes when data has to be loaded to memory. Be careful about FPS and RTS Singplayer games...where data is loaded as event triggers as you go through a map and have to deal with the stuttering.  

     

    All I wanted to make you aware (The thread starter) is that you should not rely on consumer grade equipment when building a non-consumer grade PC. It will become your weakest link and the only time you use consumer grade equipment vs something stronger is when you discover that using something stronger in a certain area will not yield any real increase in performance. 

     

  • RobgmurRobgmur Member Posts: 322

    I'm going to swap in the  mushkin 120 gig SSD and throw in the raptor HD, I don't mind the extra little $$ if it means a good setup for the future. I play both mmorpgs and shooters ect. I might also go with 1 MSI gtx 580 (the OCd one) i heard its a great card. I'll prabably only be playing on a 23-27'' single mon. soon down the road whne the prices drop and the new beefy games ccome out I'll get another 580 for SLI. I might also look at the 590 soon in the near future when it matures a little.  Do you think the 1 msi gtx 580 would run high end games now and the new releasing ones at higher ends?

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • RobgmurRobgmur Member Posts: 322

    I know i keep asking for your guys opinions a lot.. the internet out here is so horrid it takes forever to search through online reviews.

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    At 120-128 GB, most of the new SSDs aren't really any faster than the older Crucial RealSSD C300 or first generation SandForce drivers of the same capacity.  The Vertex 3 is the exception, and will probably be joined by a number of other identical drives as the usual suspects jump on the second generation SandForce bandwagon.

    But while the Vertex 3 is faster, it's benchmark fast, not necessarily real-world fast.  There will certainly be real-world situations in which the Vertex 3 would be vastly better than a Vertex 2.  If a task might have taken 10 seconds with hypothetical, infinitely fast storage (because you have to wait on the processor or whatever), then with a hard drive, maybe it takes 30 seconds, because you have to wait on the hard drive a lot.  Get a good SSD and maybe it takes 11 seconds, because you're still mostly waiting on the processor, and storage is no longer a meaningful bottleneck.  Get a better SSD and maybe you reduce that to 10.8 seconds.  The difference between 11 and 30 is a big deal.  I say the difference between 11 and 10.8 doesn't matter much.  A lot of "real world" measurements in SSD reviews find that the good SSDs are all nearly the same, even if they diverge wildly in synthetic benchmarks, but they're often dramatically faster than any hard drive.

    -----

    Don't get a VelociRaptor.  It's a waste of money.  In the above hypothetical example, maybe a typical 7200 RPM hard drive takes 30 seconds, and the fastest 7200 RPM hard drive on the market could cut that to 27 seconds.  Maybe a VelociRaptor can reduce that to 20 seconds.  Back when SSDs didn't exist, there wasn't an 11 second SSD option on the market.  For someone who wanted faster storage, the VelociRaptor was by far the fastest consumer hard drive on the market.  There were some enterprise hard drives that were faster, but those didn't use SATA, so they would take outlandish setups to use.  Against that background, the VelociRaptor makes sense.

    But now, to buy a VelociRaptor, you're paying about as much as you would for an SSD, without getting performance like an SSD.  There could be some very peculiar edge cases where a VelociRaptor makes sense, but if you have to ask, then you're not in one of those edge cases.

    The only reason to buy a hard drive at all is if you need more storage than you're willing to buy in an SSD.  If you have 1 TB of videos, pictures, music, and various junk laying around that you don't want to get rid of, then you need 1 TB of storage, but not a 1 TB SSD, as that would cost around $3000.  Personally, I don't have many GB of random junk, so I have just a 120 GB SSD and no hard drive, other than the $30 external hard drive I use for a periodic image backup of my SSD.  (Most people are terrible about backing up their data, but I'm not one of them.)

    But most of that junk doesn't need SSD speed.  You want to put real programs on an SSD, but for loading a video, it doesn't matter if it loads from a slow hard drive, as that will be plenty fast enough.  So you can get a 120 GB SSD for your programs, together with a 2 TB hard drive for random junk, and have all of the capacity you need, while also having everything be fast when speed matters.

    So the only things you put on a hard drive are those where speed doesn't matter.  If you buy a VelociRaptor, you're paying extra for more speed, precisely where it doesn't matter.  That's a waste of money.

    ------

    For monitor size, as it affects what sort of video card you need, what matters is pixels, not inches.  A 70" HDTV with a 1920x1080 resolution will put exactly the same strain on a video card as a 15" laptop with the same resolution.  The really high end video card setups really only make sense for people who are playing games at a resolution of 2560x1600 or higher.  If you're going to play games on a 1920x1080 monitor, then you might as well revert to the Asus P8P67 Pro or Gigabyte GA-P67A-UD4, get two of these in CrossFire:

    http://www.newegg.com/Product/Product.aspx?Item=N82E16814102932

    and call that good enough.  $280 in video cards, and at stock settings, it will usually outperform whichever GeForce GTX 580 you're looking at.

  • QuizzicalQuizzical Member LegendaryPosts: 22,130

    Originally posted by Robgmur

    I might also look at the 590 soon in the near future when it matures a little.

    Don't do that.  The dual GPU cards really only make sense if you want more than two GPUs.  A pair of Radeon HD 6990s makes sense for people looking to drive extremely large Eyefinity setups, but you really have to build the whole system around it.

    A GeForce GTX 590, on the other hand, makes sense for exactly no one.  It's a terrible card.  The basic problem with the GeForce GTX 590 is that it is the most failure-prone video card in a long time, possibly ever. Quite a few of them died before finishing the review process. Now, some of those were overclocked when they died, but comparable or larger overclocks usually don't kill other cards, and especially not that quickly. For example, this got passed around a lot:



    http://www.youtube.com/watch?v=sRo-1VFMcbc



    If you want to know how they got the video of that, they had two GTX 590s. One unexpectedly died at a given overclock. So they said, let's try that with the other, and take a video. Surprise, the other one died, too, and that's how they got the video. They only added about 0.1 V above the stock voltage, and very, very few cards will die quickly if you only increase the voltage by 0.1 V. Not many will die that quickly if you add 0.2 V to the stock voltage, for that matter, though that's a serious risk to fry the card eventually.



    There are two reasons for it. One is that the cooler on the card is woefully inadequate. See here, for example:



    http://www.hardware.fr/articles/825-4/dossier-nvidia-repond-amd-avec-geforce-gtx-590.html



    Even if you can't read French, the pictures are pretty self-explanatory. Usually the hottest part of the interior of a video card is the GPU chip, which usually doesn't go much past 90 C. Here, most of the interior of the card is over 90 C, and the peak is 111.9 C. That is flatly unsafe. Even the highest end capacitors that go into consumer electronics are only rated for 105 C, and I doubt that there are VRMs or MOSFETs built to handle 110+ C, either. Maybe that's why a number of the dead GTX 590s had exploding MOSFETs as the culprit.



    The real problem is that the card puts out too much heat. Nvidia claims a TDP of 365 W, but that's a complete lie. An honest TDP would probably be more like 450 W. And no one knows how to properly cool 450 W on air in a standard length two slot cooler. Not Nvidia, not AMD, not Intel, not Zalman, not Arctic Cooling, no one.



    And then Nvidia went too cheap on power circuitry. They have power circuitry that would probably be more or less safe at 375 W. But that's not enough for a 450 W card.



    Nvidia doesn't have a good way to cap power draw, either. The Radeon HD 6990 has PowerTune, where the card can measure power consumption in real time. If it goes over 375 W, then within a fraction of a second, the card will reduce clock speeds by just enough to decrease power consumption to 375 W. As soon as power consumption is back down, the clock speeds will return to normal.



     Nvidia doesn't have that. Nvidia drivers do check for programs known to push cards especially hard, and reduce clock speeds when those programs are running. FurMark and OCCT are the best known such programs. But if some new game launches in which one particular part pushes the cards harder than expected, if you run that game before Nvidia finds out and issues a driver update, or even if the update is already out but you don't have it, you could end up with a dead card. The title screen of StarCraft II did that to quite a few people.



    Now, most cards can over-engineer the parts somewhat. If a card is supposed to be safe at 150 W, then you can design the card such that 200 W probably won't kill it, and then if a game does spike power consumption up to 200 W, you're all right. But Nvidia did not do that with the power circuitry on the GeForce GTX 590, and could not do it on the cooling system.



    The other thing that Nvidia cards can do is measure temperatures, and if temperatures get too hot, reduce clock speeds dramatically to try to save the card. But that means not compensating until after the card has overheated, and by then, damage might already be done.



    AMD's PowerTune doesn't have to know what programs you're going to run, or how hard they push the card. If some power virus that AMD has never seen before starts running, within a fraction of a second, the card will recognize that something is amiss and cut clock speeds. An Nvidia card won't do anything about it until the card has already overheated. AMD's PowerTune can decrease clock speeds by just enough, too, so if a game pushes a card a little too hard, and 830 MHz will make it run too hot, it can briefly cut the clock speed to 800 MHz, and you'll never know it happened. Temperature-based clock speed reductions are far more dramatic than that, as the card decides it needs to nearly stop putting out heat right that instant, and your frame rates get reduced to a slide show if they kick in.



    Nvidia will try to tell you that driver updates fixed the problem. Don't believe them. Driver updates can't fix hardware problems. All that driver updates can do is to make the fan run faster (in which case, you lose the noise advantage over a Radeon HD 6990) or turn the clock speeds slower (in which case, it's not in the Radeon HD 6990's league in performance any more). Reports say that it mostly did the latter, with newer drivers far quicker to turn down clock speeds. So don't expect to get the performance that you might have seen in GeForce GTX 590 reviews.



    The GeForce GTX 590 isn't the only card prone to overheat, of course. The reference versions of the GeForce GTX 480, GeForce GTX 470, and Radeon HD 5970 had their own problems with heat. But the GeForce GTX 590 is much, much worse about it.



     Furthermore, there are better alternatives to the GeForce GTX 590. AMD's own Radeon HD 6990 will tend to perform better than a GTX 590, and while it does have heat and power issues, those can be addressed by building a system around the card. You simply can't do that with a GeForce GTX 590. But dual GPU cards only make sense if you're going to use more than two GPUs.  Otherwise, you can just get two single GPU cards in CrossFire or SLI.

  • RobgmurRobgmur Member Posts: 322

    I edited the original post with the new set-up. I decided to go with SLI MSI 580s. Will the GIGABYTE GA-P67A-UD3P-B3 LGA 1155 Intel P67 mother board work well for future proof? or should i go with something more

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • ShinamiShinami Member UncommonPosts: 825

    The problem with the Dual GPU cards is the fact that the PCI-EXpress slot itself has a limit in its specification. 

     

    When you have two video cards, you have at least two PCI Express Slots that have their own independent bandwtih. Its actually less strain on the system because you are literally connected at two points drawing half of what is needed compared to having ONE point that is truly drawing a lot and putting a strain on the system.

     

    I am an Nvidia Fan, but this is actually one of the video cards I do not recommend anyone into buying. The same is true for 6990s. Its not that EITHER CARD is horrible, but the strain it does on the motherboard and even power supply. When the draw exceeds a certain point, voltages affect a system or device a lot greater. This affect is mostly going to be FELT by the SLOT itself...

     

    Now, when they make a stronger PCI-Express SLOT that has higher integrity or a new generation, along with cards under a new architecture that is not just a dual GPU but a dual GPU with a reason and incentive to buy over a single GPU solution, then by all means I would recommend it to anyone. 

     

  • ShinamiShinami Member UncommonPosts: 825

    Originally posted by Quizzical

    At 120-128 GB, most of the new SSDs aren't really any faster than the older Crucial RealSSD C300 or first generation SandForce drivers of the same capacity.  The Vertex 3 is the exception, and will probably be joined by a number of other identical drives as the usual suspects jump on the second generation SandForce bandwagon.

    But while the Vertex 3 is faster, it's benchmark fast, not necessarily real-world fast.  There will certainly be real-world situations in which the Vertex 3 would be vastly better than a Vertex 2.  If a task might have taken 10 seconds with hypothetical, infinitely fast storage (because you have to wait on the processor or whatever), then with a hard drive, maybe it takes 30 seconds, because you have to wait on the hard drive a lot.  Get a good SSD and maybe it takes 11 seconds, because you're still mostly waiting on the processor, and storage is no longer a meaningful bottleneck.  Get a better SSD and maybe you reduce that to 10.8 seconds.  The difference between 11 and 30 is a big deal.  I say the difference between 11 and 10.8 doesn't matter much.  A lot of "real world" measurements in SSD reviews find that the good SSDs are all nearly the same, even if they diverge wildly in synthetic benchmarks, but they're often dramatically faster than any hard drive.

    -----

    Don't get a VelociRaptor.  It's a waste of money.  In the above hypothetical example, maybe a typical 7200 RPM hard drive takes 30 seconds, and the fastest 7200 RPM hard drive on the market could cut that to 27 seconds.  Maybe a VelociRaptor can reduce that to 20 seconds.  Back when SSDs didn't exist, there wasn't an 11 second SSD option on the market.  For someone who wanted faster storage, the VelociRaptor was by far the fastest consumer hard drive on the market.  There were some enterprise hard drives that were faster, but those didn't use SATA, so they would take outlandish setups to use.  Against that background, the VelociRaptor makes sense.

    But now, to buy a VelociRaptor, you're paying about as much as you would for an SSD, without getting performance like an SSD.  There could be some very peculiar edge cases where a VelociRaptor makes sense, but if you have to ask, then you're not in one of those edge cases.

    The only reason to buy a hard drive at all is if you need more storage than you're willing to buy in an SSD.  If you have 1 TB of videos, pictures, music, and various junk laying around that you don't want to get rid of, then you need 1 TB of storage, but not a 1 TB SSD, as that would cost around $3000.  Personally, I don't have many GB of random junk, so I have just a 120 GB SSD and no hard drive, other than the $30 external hard drive I use for a periodic image backup of my SSD.  (Most people are terrible about backing up their data, but I'm not one of them.)

    But most of that junk doesn't need SSD speed.  You want to put real programs on an SSD, but for loading a video, it doesn't matter if it loads from a slow hard drive, as that will be plenty fast enough.  So you can get a 120 GB SSD for your programs, together with a 2 TB hard drive for random junk, and have all of the capacity you need, while also having everything be fast when speed matters.

    So the only things you put on a hard drive are those where speed doesn't matter.  If you buy a VelociRaptor, you're paying extra for more speed, precisely where it doesn't matter.  That's a waste of money.

    ------

    For monitor size, as it affects what sort of video card you need, what matters is pixels, not inches.  A 70" HDTV with a 1920x1080 resolution will put exactly the same strain on a video card as a 15" laptop with the same resolution.  The really high end video card setups really only make sense for people who are playing games at a resolution of 2560x1600 or higher.  If you're going to play games on a 1920x1080 monitor, then you might as well revert to the Asus P8P67 Pro or Gigabyte GA-P67A-UD4, get two of these in CrossFire:

    http://www.newegg.com/Product/Product.aspx?Item=N82E16814102932

    and call that good enough.  $280 in video cards, and at stock settings, it will usually outperform whichever GeForce GTX 580 you're looking at.

    Interesting post....however

     

    "In the above hypthetical example....."

     

    Perhaps you should edit your post to read

     

    "I do not agree with the above hypothetical example because I read it from a web site and know from "reading" that this poster, who has a masters degree in computer sciences with a concentration in operating system design and network programming, who spend over 10 years in the industry and is an enthusiast owns the hardware in question and has thoroughly tested it....must be wrong because this poster has always gone against the claims made by the top 10 web sites over and over again, considerting that only two people who created the top 10 sites on the internet have Degrees in computer sciences and most of their time is spent editing a web page and talking about the latest smart phones while this poster actually works testing, programming and writing technical documentation" 

     

    Please don't insult me by calling my "ideas" as Hypothetical when I thorougly have tested them. I don't come around forumboards asking you to prove your mathematical background. I do "frown" upon "opinions" being posted in a Hardware forum where the subject of questions tend to call for Objective Analysis rather than Subjective Analysis.

     

    Its amazing I used a four year Raptor HDD and was able to have smooth playing, then as soon as I used a SATA III 7200 RPM HDD, while the transfer speed was higher from HDD to HDD...I noticed stuttering and slight annoyances in gameplay to some games that really did get to me. 

     

    Of course you like mathematics and believe me I am a fan of mathematics (Discrete Mathematics, Linear Algebra, Abstract Algebra, Number Theory and Differential Equations are my favorite maths). So lets go into Arithmetic.....How about 10000/7200 = 1.3888-> In short, a 10K RPM HDD spins 38.88% faster than a 7200 RPM HDD. A velociraptor is a 2.5 inch HDD which has a GIANT HEATSINK as a housing. It has smaller and denser platters...and it does perform a lot better..

     

    I AM NOT refering to LOADING TIMES when you load a file like loading a LEVEL in CRYSIS or loading a map in an MMORPG when you first LOG IN or watching a "NOW LOADING" screen which of course you load them faster on the 10K RPM but does it warrant buying one? 

     

    It does when you think about "Micro-loading" where the game needs to LOAD FILES IN REAL TIME after actually entering the game and within the very same map like event triggers, enemy spawn data and other things. I REFER TO THE TIME IT TAKES TO LOAD FILES IN REAL TIME TO THE POINT stuttering and stalls go away because those files can be FOUND AND LOADED without slowdowns occuring. This is the MAIN reason you would buy a 10K RPM HDD For gaming vs a 7200 RPM HDD...as well as the actual reliability behind the drive. 

     

    The Second point I make about  7200 RPM HDDs is capacity. Sure, they have a LOT HIGHER CAPACITY. We can all nod our faces in agreement at finding drives in the TB range...However that comes wtih a price. Fill up a 7200 RPM HDD and as LOW as 33% you start experiencing the drive starting to slow down. Eventually you reach a point where a drive at around 50 - 75% slows down so much that you literally get the performance of a 5400 RPM HDD. A 10K RPM HDD has a much higher threshold to the point you can fill up a drive to 50 - 75% and the impact on performance is not as LARGE as what you see on a 7200 RPM HDD when it filles up. THIS IS THE SECOND REASON WHY I CHOOSE 10K RPM HDDs vs 7200 RPM HDDs.

     

    By the way, don't take this as an insult (A lot of mathematicians where I am from feel insulted if you ask them if they know certain equations), Do you like Differential Equations? :) 

     

    Consider this like a "peace" as I don't want to throw some high-insult post. 

     

    http://www.earsc.com/HOME-Electronic/engineering/TechnicalReference/TechnicalWhitePapers/IsolationTechniquesfor2-5HDD/index.asp?SID=557

     

    Do you like differential equations? :) I like math at work. These deal with sound and dampening, considering these 2.5 inch HDDs (Velociraptors as 2.5 inch and make tons of noise)...

     

    Like me or hate me.... You can share mathematics with me and I would understand them. My only REAL COMPLAINT to 10K RPM HDDs is the noise they make. The thing that pisses me of the MOST in the industry today is the fact that an SSD itself has the potential to lose Magnitudes of Power between 2 and 4 and really hope that a better design comes around in a few years to minimize the issue. This is why I dont rely on them for gaming. 

     

    I do suppose you could buy an SSD for gaming and use it.....but then thanks to the way they work, you would deal with losing such magnitudes of power and the SSD actually will LAST a lot less than a Raptor. If you lose four magnitudes of power from a 15 million MTBF, you go from 1.5 x 10^7 to 1.5 x 10^3, which is a measly 1,500 hours (Remember all the failures which occured with earlier SSDs that were used heavily for gaming?), Combine the fact of losing half the lifespan of an object when it operates 10C above tolerance and that drops to around 750 hours under the worse case scenario or 7.5 x 10^2. 

     

    The AVERAGE LOSS of Magnitudes of Power tends to be Two these days, sometimes three....If it were two, it would be 75,000 hours (with 10C above tolerance WHICH WILL HAPPEN if you game with a SSD) or 7.5 x 10^4 which is not so bad unless of course we deal with structural damage of any kind. 

     

    Well, I'll go now. Good Luck. ^_^ 

  • RobgmurRobgmur Member Posts: 322

    Would you think sub in ..say a ASUS P8P67 WS REVOLUTION LGA 1155 Intel P67  and stick with the SLI, or lose the SLI then and go with a single (insert recommended future proof Video card here) maybe the single msi 580gtx? Or .. Or grab another SLI set-up.

    Also, would you think losing the SSD and go for a raptor for a reliable set-up? Plus I would save quite a bit $$ and lose minimul preformance. 'According to what you were saying' ?

    My main concern was ..say running Battlefield 3 and Skyrim at  launch on MAX 60-80 FPS would be nice but we don't know how the games will code out.. In that ball park would be great though. If you have any swap-out suggestions.. I'm all ears.

    I selected the MSI GTX 580s because I read many great reviews on them. Yes, the price is a little steep for their grade but from my understanding it was worth it.

    *Corsair Obsidian Series 650D *i5-2500K OC'd ~ 4.5
    *Asus P8Z68-V PRO/GEN3 mother board
    * Radeon HD 7970
    *8GB (4GBx2) 1600MHz Kingston HyperX
    *240GB Corsair Force GT Series SATA-III SSD

  • AntariousAntarious Member UncommonPosts: 2,809

    Originally posted by Robgmur

    Would you think sub in ..say a ASUS P8P67 WS REVOLUTION LGA 1155 Intel P67  and stick with the SLI, or lose the SLI then and go with a single (insert recommended future proof Video card here) maybe the single msi 580gtx? Or .. Or grab another SLI set-up.

    Also, would you think losing the SSD and go for a raptor for a reliable set-up? Plus I would save quite a bit $$ and lose minimul preformance. 'According to what you were saying' ?

    My main concern was ..say running Battlefield 3 and Skyrim at  launch on MAX 60-80 FPS would be nice but we don't know how the games will code out.. In that ball park would be great though. If you have any swap-out suggestions.. I'm all ears.

    I selected the MSI GTX 580s because I read many great reviews on them. Yes, the price is a little steep for their grade but from my understanding it was worth it.

     

    In regards to your SSD question.   There is a lot of information on SSD drives available and it would do you well to read it.  Simply because if you are interested in a technology you should understand it and its limitations.   As opposed to a theoretical situation someone presents to you.   User education is always the best so you can make an informed decision.   SSD is going to depend entirely on how "you" use your system.   If you are someone that tends to uninstall software often, you may not want to use SSD.   Simply because the biggest weakness to SSD is writing data to the same areas over and over.   This is going to be the biggest factor that impacts the life of your SSD.  

     

    You can also set up a smaller SSD as your OS drive and then use another non-ssd for data (aka game installs etc).   This is very common and in fact is what I use personally.   Tho I do have a few games on my SSD's in all 3 systems that have them.

     

    I don't really have a recommendation on your vid card choice.   Best I have is if you can afford the 580 and you like what you've read then that's your choice.   SLI to me is mostly the cost involved, making sure you have enough space between your cards, the heat involved, noise and power consumption.   As compared to the actual benefit of running SLI.

     

    *edit for clarity*

     

    on the vid card when I say "that's your choice" its to mean this is probably the card you should buy.   I have a 570gtx in this paticular machine and its ok... but after I got it I read about some build issues showing up in 570's that didn't affect 580's.   So now if I could go back I'd have ordered the 580 because it fit in my budget I just wanted to save some money on this build.

Sign In or Register to comment.