If you don't care if your computer is slow, then sure, you don't need an SSD. But then, if you don't care if the computer is slow, then there are lots of other places that you could cut way back, too.
The question of whether to get an SSD or not is basically one of what you want your computer to do when you tell it to do something:
1) wait, wait, wait, and then eventually do what you said, or
2) do what you said promptly, rather than making you wait.
An SSD is the second option. You probably have some idea of when your current computer makes you wait. Furthermore, you probably have some idea of whether that waiting is due to downloading something from the Internet. When it isn't a download, an SSD would mean you have to wait a lot less.
SSD is in no way necessary at all, especially for someone having to make a gaming rig on a budget.
He would be WAYYYYYYYYYYYYYYYYYYYYYYYYYYYYY better off investing the money into.....
1. GPU
2. CPU
3. PSU or motherboard.
SSD is a luxury item right now. On average it will load windows 7-15 seconds faster and shave off a few seconds for most programs. Nothing major whatsoever and really not that important for gaming.
My build already has a fairly high end CPU, PSU, and motherboard. There's no real sense in spending more on any of those. Now, if someone were going with a cheap junk power supply or motherboard, then yes, I'd agree that getting something decent for those parts was a higher priority than the SSD.
For the video card, you can make more of an argument. But a Radeon HD 6870 is already a very capable card. Video cards are also easy to upgrade two or three years down the road--and a lot easier than trying to move your OS onto an SSD later.
"I would take a NVIDIA card over an ATI card any day of the week. I have used ATI almost exclusively and recently switched and have fewer problems and they bring drivers out faster."
Here comes the FUD, I see. Nvidia and AMD both have pretty good drivers. Cards branded as ATI are mostly off the market now, except for the low end.
The difference between the 990FX and 970 chipsets is PCI Express bandwidth. 990FX is good for CrossFire or SLI, but there's no need for it for a single video card.
Why are you still looking to spend $9 on thermal paste, rather than using what comes with the heatsink? The difference that makes will basically be a rounding error. And why a cheap aftermarket heatsink? Is that better than the stock heatsink that AMD ships? It might be, or it might not be, and even if it is better, it isn't much better.
Why are you still looking to pay $90 for 8 GB of system memory? There are plenty of options for that capacity under $50.
But the power supply is the real problem. A Corsair GS series power supply is all right. A Coolmax anything probably isn't. I don't know how bad that particular model is, but surely you don't, either. Coolmax has made enough power supplies that are complete garbage that I'd assume that all Coolmax power supplies are junk until proven otherwise. And don't hold your breath waiting for it to be proven otherwise. It's probably merely mediocre rather than completely awful, as 80 PLUS certification does tend to filter out the really bad ones, but I wouldn't recommend it even if it were cheap. And you're paying more for it than the high end power supply that I linked in my build.
The difference between the 990FX and 970 chipsets is PCI Express bandwidth. 990FX is good for CrossFire or SLI, but there's no need for it for a single video card.
990x/FX have 5200 HT link speed vs 4800 for 970. Is good for 8150 which uses HT for core to core communications. I'd even bump up the mobo to the Extreme 4 that has 8+1 power vs 4+1 for the extreme 3
Why are you still looking to spend $9 on thermal paste, rather than using what comes with the heatsink? The difference that makes will basically be a rounding error. And why a cheap aftermarket heatsink? Is that better than the stock heatsink that AMD ships? It might be, or it might not be, and even if it is better, it isn't much better.
If you read the Thermal grease reviews you would know that Mx2 paste is much better than stock grease. Go read the thermal grease shootout. And that HS with 2 fans in a push/pull will put a H70 to shame. And the HS you linked has abysimal height making it block ram slots.
Why are you still looking to pay $90 for 8 GB of system memory? There are plenty of options for that capacity under $50.
Yes but with an unlocked processor and a mobo that supports up to 2100 why not get the faster ram. it does make a difference.
But the power supply is the real problem. A Corsair GS series power supply is all right. A Coolmax anything probably isn't. I don't know how bad that particular model is, but surely you don't, either. Coolmax has made enough power supplies that are complete garbage that I'd assume that all Coolmax power supplies are junk until proven otherwise. And don't hold your breath waiting for it to be proven otherwise. It's probably merely mediocre rather than completely awful, as 80 PLUS certification does tend to filter out the really bad ones, but I wouldn't recommend it even if it were cheap. And you're paying more for it than the high end power supply that I linked in my build.
That 850W power supply was checked out by Proclockers, 100A, good temps and build quality. And it's Modular.
As I said befor on the processor FX8150 in a double blind study felt smoother than a 2700K. Bash the study all you want, it was supervised, there were no AMD banners or tricks. IT IS A BETTER PROCESSOR.
My setup will get him a massive overclock, a faster SSD, and 100$ headroom to upgrade what he sees fit. I'd bump up the Mobo to the Asrock 990fx extreme 4, Or get a higher grade 6870 that has a better cooler.
"990x/FX have 5200 HT link speed vs 4800 for 970. Is good for 8150 which uses HT for core to core communications. I'd even bump up the mobo to the Extreme 4 that has 8+1 power vs 4+1 for the extreme 3"
For a single video card, the extra HyperTransport bandwidth doesn't matter, as you can't use it anyway. I'd be utterly shocked if Zambezi has one core communicate with another by routing the signal through the southbridge rather than merely across the CPU. I'd like to see a reference on that. I can understand wanting more capable power delivery, but does the original poster strike you as a serious overclocker?
"If you read the Thermal grease reviews you would know that Mx2 paste is much better than stock grease. Go read the thermal grease shootout. And that HS with 2 fans in a push/pull will put a H70 to shame. And the HS you linked has abysimal height making it block ram slots."
And those tests were conducted in a way that will produce larger differences than most people. Note that most of the ones down at the bottom are various household substances for which they said, let's try using this as thermal paste and see what happens.
"Yes but with an unlocked processor and a mobo that supports up to 2100 why not get the faster ram. it does make a difference."
Never mind what the motherboard supports. The memory controller only supports 1866 MHz. And you can get 8 GB of 1866 MHz DDR3 for $60. The real-world performance difference between 1866 MHz memory and 1600 MHz won't often be more than a rounding error anyway.
"That 850W power supply was checked out by Proclockers, 100A, good temps and build quality. And it's Modular."
I'm guessing you're talking about this pathetic excuse for a power supply review.
Look at the hardware they used. They might have pulled 300 W from it, if that. Certainly not even 400 W. That's no way to test a 1000 W power supply. They didn't open it up to inspect the build quality, either.
For comparison, the Rosewill Capstone was tested by Jonny Guru.
"990x/FX have 5200 HT link speed vs 4800 for 970. Is good for 8150 which uses HT for core to core communications. I'd even bump up the mobo to the Extreme 4 that has 8+1 power vs 4+1 for the extreme 3"
For a single video card, the extra HyperTransport bandwidth doesn't matter, as you can't use it anyway. I'd be utterly shocked if Zambezi has one core communicate with another by routing the signal through the southbridge rather than merely across the CPU. I'd like to see a reference on that. I can understand wanting more capable power delivery, but does the original poster strike you as a serious overclocker?
The HT link speed affects more than just the HT links. it's the data speed of the EIB also..I think. I'd rather give someone the tools they need to get into overclocking. It cant hurt
"If you read the Thermal grease reviews you would know that Mx2 paste is much better than stock grease. Go read the thermal grease shootout. And that HS with 2 fans in a push/pull will put a H70 to shame. And the HS you linked has abysimal height making it block ram slots."
And those tests were conducted in a way that will produce larger differences than most people. Note that most of the ones down at the bottom are various household substances for which they said, let's try using this as thermal paste and see what happens.
"Yes but with an unlocked processor and a mobo that supports up to 2100 why not get the faster ram. it does make a difference."
Never mind what the motherboard supports. The memory controller only supports 1866 MHz. And you can get 8 GB of 1866 MHz DDR3 for $60. The real-world performance difference between 1866 MHz memory and 1600 MHz won't often be more than a rounding error anyway.
The controller goes higher, I have my PII X6 at 2000 atm.
Not so, look at the benchies.....Cryptograpy gets a boost, 3dmar11, BF3, CivV, MoH. Honestly I agree 2000 vs 1866 or even 1600 is minimal. 1866 would give good compromize and mabye some.
"That 850W power supply was checked out by Proclockers, 100A, good temps and build quality. And it's Modular."
I'm guessing you're talking about this pathetic excuse for a power supply review.
Look at the hardware they used. They might have pulled 300 W from it, if that. Certainly not even 400 W. That's no way to test a 1000 W power supply. They didn't open it up to inspect the build quality, either.
For comparison, the Rosewill Capstone was tested by Jonny Guru.
Originally posted by grndzro "If you read the Thermal grease reviews you would know that Mx2 paste is much better than stock grease.
.....
You need to read this. IMO 5 degrees isn't chump change. It's probably less of a difference but applying it to CPU and Video card can help a lot.
I just read it.
The difference between Arctic Silver 5 (widely regarded as one of the best, and tied for 2nd on this web site) and Pink Lipstick was 8C.
I couldn't find anything called MX2 on this list. There was an Arctic Cooling MX-4, which performed pretty mediocrely (and tied with Mayonnaise, the popular condiment, fwiw).
Most every single ~real~ heatsink compound (that is, those marketed for that specific purpose) performed within a 3C band, with none outside of a 5C band, which could very well be argued to be within the margin of error of the test. With one exception: any compound Rosewill brand performed particularly poorly.
Aside from that, I think the only real lesson to take away from this chart is: Use some kind of heatsink compound, it doesn't really matter what though, they are all about the same. If you work on a lot of computers, and want to spring for a tube of the good stuff, or really want that extra edge for very high and tight overclocks, then by all means. But for 99% of people who just want to put together a computer and have it work well, or even with light or modest overclocks, the stock stuff that comes with the heat sink is more than good enough. Heck, even Mayonnaise is good enough...
I will agree that SSD is totally unnecessary. It helps, but not so much that it is anywhere approaching worth the additional cost. If money is an object, there is so much else you can spend it on that will help you a great deal more.
But, people are giving you great options for putting your computer together, which is great, but remember that you have to put this together. If you friend can help you, that would be ideal. It isn't terribly hard, but there are something you could do that will destroy your machine before you even turn it on, as a result of static electricity, among other things.
It isn't hard, by any stretch, but you would be better off with some help for your first time out.
"There are two great powers, and they've been fighting since time began. Every advance in human life, every scrap of knowledge and wisdom and decency we have has been torn by one side from the teeth of the other. Every little increase in human freedom has been fought over ferociously between those who want us to know more and be wiser and stronger, and those who want us to obey and be humble and submit."
John Parry, to his son Will; "The Subtle Knife," by Phillip Pullman
"You need to read this. IMO 5 degrees isn't chump change. It's probably less of a difference but applying it to CPU and Video card can help a lot."
It looks like we basically agree that different thermal paste does make a relatively small difference, but disagree over just how big of a difference it is and how much that matters.
51 of the 55 thermal pastes listed there (I'm excluding the various household substances) come in somewhere between 33 and 38 C. That's nearly all of them in a band only 5 C across.
What do you think would happen if you picked one thermal paste and tested it 55 times? Think it would give exactly the same measurement every time? It most certainly would not. There's some variation in room temperature, and in the measured CPU temperature. There's also variation in how the thermal paste is applied, and that can make a difference of another degree or two. The analogous statement of all but a few outliers being in a particular band would probably require a band 3 or 4 C across.
Now yes, 5 C across is bigger than 3 or 4 C. So a better thermal paste probably does tend to make a difference of a couple of degrees or so. Now does a couple of degrees matter? If you're going for a heavy overclock and looking to fry the CPU, then yes, it does. At stock speeds or a moderate overclock, no it doesn't.
"The controller goes higher, I have my PII X6 at 2000 atm.
Not so, look at the benchies.....Cryptograpy gets a boost, 3dmar11, BF3, CivV, MoH. Honestly I agree 2000 vs 1866 or even 1600 is minimal. 1866 would give good compromize and mabye some."
The Zambezi memory controller is only rated at 1866 MHz. Now, you can overclock it, but overclocking a memory controller with a rated speed that is already fairly aggressive (1866 MHz is the highest ever for any CPU, and by a large margin) is a rather different matter from overclocking a memory controller for which the company went curiously conservative on the rating (after Lynnfield was rated at 1600 MHz, do you really believe Sandy Bridge can't do the same?).
And look at your link. They have a confounding problem of not merely using higher clock speeds, but also having tighter memory latencies at the higher clock speeds. So it's entirely possible that even if there were improvements, they would entirely be due to tighter latencies and not the higher clock speed.
And then look at the numbers. Dirt 3 does best at 1866 MHz, and then 1600 MHz, and then 2000 MHz only after those. Crysis 2 does best at 1600 MHz, and performs better there than at 1866 MHz or 2000 MHz. Now, the differences are rounding errors, so it's not like 2000 MHz is outright worse. But it's not better, either, and there's no sense in paying an extra $60 for a rounding error.
Originally posted by grndzro It's there 5 from the top in excellent performance section
I don't see any "Excellent Performance" section. Are you looking at the same link that you provided? Maybe you should post this link because it's certainly not the Hardware Secrets Thermal Compound Round-Up that the rest of us are looking at.
This link has 1 page of results, an unsorted chart, and a sorted bar graph.
5 from the top in the unsorted chart is Evercool EX420-TU15, which performed at a mediocre 35C (about average). It ties for 4th place in the ordered chart 5 from the top in the ordered graph is Enermax (Dow Corning TC-5121), which being in 5th place, performed at 33C, which makes it tie for second place.
THe difference between 1 and 8th place are all separated by 1C. Between 8 and 9 is 2C. 9th to 10th is 1C again. Then 10->11 is 19C (no heatsink compound at all), and in last place was chocolate, which was 43C away from 1st place.
This link is over a year old now, but it's pretty applicable.
Tom's takes a Lynnfield i7 and runs 4 different memory configurations on it, from DDR3 800 to DDR3 1600, and each of those speeds with a couple of different latencies.
The jump from DDR3 800 6-6-6 (their baseline) to DDR3 1600 8-8-8 (the highest performance they tested) was 5%. That's doubling the speed.
Their conclusion: If you can get cheaper RAM for basically the same price, go ahead, it doesn't hurt. But if you are paying so much that you could use that money to upgrade your CPU or GPU, then your wasting your money, because CPU/GPU will yield dramatically better results.
The performance gains are a bit more than a rounding error, but it isn't much. Some programs were more sensitive to it than others, and some systems are more sensitive than others (those that rely on integrated graphics, such as Llano). Some programs showed almost no difference (likely bound by other criteria). SiSoft's various encryption benchmarks showed between 0 and 0.5% difference from the lowest to highest memory speeds tested. The same memory benchmarks showed 178% increase (which is to be expected really, the clock speed has doubled).
For the games they tested, Left 4 Dead showed a decent improvement, about 8%, but it was far and away the largest delta. Interestingly enough, in some tests, higher rated memory performed poorer (DDR1300 10-10-10 being faster by 1.9% over DDR1300 7-7-7 in the GTA4 test, for instance, when they are the same chips just at different BIOS settings) - which just goes to show the margin of error in the testing.
My Overdrive benchmark jumped from 7000 to 7500 points by using ddr3 1750 8-9-8-25 timings over ddr3 1600 7-8-7-24.
It depends on your processor and how much it needs memory bandwidth. My phenom X6 @4ghz was clearly starved for bandwidth at 1600 so jumping up the memory speed yielded very good results. For a quad core jumping up from 1600 isn't a big improvment.
I would assume an 8 core needs the bandwidth even more than a 6 core so using FX8150 with DDR3 1600 is really stupid IMO.
I will get AMD optimized DDR3-2000 with really tight timings when I upgrade to Piledriver next year. I'v heard all the opinions that it makes little difference but when I open a browser with 20 tabs and it just snaps open and loads much faster there isn't much that can sway the real world performance I see.
Every one here has his or her opinion so i'll give you mine too.
SSD drives not neccessary they don't really add that much to the game exp to justify the cost if you ask me and they are really small.
You do not really need a top shelf CPU either Games are not really pushing the envelop of an 8 core system yet and OSs don't either.
I will tell you what I have and why
I started by choseing my chipset. The chipset is what will determine what features you have and can control in your bios settings like overclocking, variable fan speeds, and lots more
Chipset - AMD 790Gx
Socket is also important it will determine which processors you can fit to your MB
GA-MA790GP-DS4H (AM2+)- This is a really solid MB with a huge range of CPUs that fit it. About 3 years ago it was (not sure if they are still available) This is an old MB and its biggest detractor is RAM speeds they are slow DDR2 is old tech. This would be one very good reason to move to a newer MB $150
CPU: I started with a duel core 7750 black edition and later upgraded.(great for overclocking and easy with the chipset features in the bios) think this one ran about $89 back then
NewerCPU $179
vendor
CPU Model
Frequency
L2
Cache
L3
Cache
Core Name
Process
Stepping
Wattage
System
Bus(MT/s)
Since BIOS Version
AMD
Phenom II X6 1100T
3300MHz
512KBx6
6MB
Thuban
45nm
E0
125W
4000
F7I
RAM: Dual CHannel DDR2 1066, the MB can support up to 16gig of RAM but I've never found 4gig sticks in this flavor so I stuck with 8 gig. This old tech for memory and would be a good reason to look at some of the newer MBs and chipsets.
Vid card: this is where opinions vary massively. Any vid card that can run between 30 and 60fps is all you'll ever need and you can get this for less than $200.
This is not a top of the line best of everything but it is a solid system I built rought 3 years ago and it will be good for about another 3 years if i have anything to say
I am assuming you will reuse some of your old gear like the CD, mouse, keyboard, monitor, maybe even a fan or two.
This system provided you can get all the parts will run you about $1168
AMD Phenum II x4 3.6Ghz 975 black edition 8 gig Ram Nvidia GeForce GTX 760
GA-MA790GP-DS4H (AM2+)- This is a really solid MB with a huge range of CPUs that fit it. About 3 years ago it was (not sure if they are still available) $150
Socket AM2+ has been obsolete for about three years, and does not support modern processors or even modern DDR3 memory. That doesn't mean you need to scrap it right this minute if you already have it, but it does mean that you shouldn't buy it new.
Thuban processors from around April 2010 can go into some socket AM2+ motherboards, but even those seem to be discontinued and disappearing. Zambezi processors are meant for Socket AM3+, and while they can go into some AM3 motherboards, they can't go back to AM2+, as it doesn't have a DDR2 memory controller. Llano processors are Socket FM1, which isn't backward compatible to anything before it. Obviously, Intel processors wouldn't go into an AMD motherboard, either.
GA-MA790GP-DS4H (AM2+)- This is a really solid MB with a huge range of CPUs that fit it. About 3 years ago it was (not sure if they are still available) $150
Socket AM2+ has been obsolete for about three years, and does not support modern processors or even modern DDR3 memory. That doesn't mean you need to scrap it right this minute if you already have it, but it does mean that you shouldn't buy it new.
Thuban processors from around April 2010 can go into some socket AM2+ motherboards, but even those seem to be discontinued and disappearing. Zambezi processors are meant for Socket AM3+, and while they can go into some AM3 motherboards, they can't go back to AM2+, as it doesn't have a DDR2 memory controller. Llano processors are Socket FM1, which isn't backward compatible to anything before it. Obviously, Intel processors wouldn't go into an AMD motherboard, either.
I doubt you can get this stuff now as I mention it is a 3 year old sysem and it is what I run atm. I am simply demonstraiting that my 3 year old system is still in the running and will be for at least a few more years before it is totally obsolete.
I am going to build a new system in the near future, I jsut have not finished researching the parts I want to put in it.
AMD Phenum II x4 3.6Ghz 975 black edition 8 gig Ram Nvidia GeForce GTX 760
Comments
Got it all down to 1106$ using a combo of newegg, superbiiz and Tigerdirect
Motherboard is bumped up to Asrock 990FX too for the 5.2 hypertransport, Power supply bumped up to 850 from 800.
Processor Superbiiz FX-8150 255$
Mobo Newegg Asrock 990FX 130$
Ram Tigerdirect Transcend 8GB DDR3-2000 90$
Heatsink Superbiiz Cooler Master Hyper TX3 18.49$
Video card TigerDirect HIS H687FN1GD Radeon HD 6870 150$
Power Supply Tigerdirect Coolmax ZP-850B ZP Series Power Supply - 850W 95$
OS Superbiiz windows 7 HP 99$
Case TigerDirect Ultra BLACK Edition Full-Tower ATX/mATX 80$
Hard drives Newegg 320GB WD & 64GB OCZ Agility 3 SSD 60$ and 90$
Thermal Paste Newegg AC-MX2 9$
CD/DVD Newegg LITE-ON Black 18X DVD-ROM 48X CD-ROM 19$
Extra fan for CPU heatsink Newegg ARCTIC COOLING AF9 PWM 92mm Case Fan 10$
My build already has a fairly high end CPU, PSU, and motherboard. There's no real sense in spending more on any of those. Now, if someone were going with a cheap junk power supply or motherboard, then yes, I'd agree that getting something decent for those parts was a higher priority than the SSD.
For the video card, you can make more of an argument. But a Radeon HD 6870 is already a very capable card. Video cards are also easy to upgrade two or three years down the road--and a lot easier than trying to move your OS onto an SSD later.
"I would take a NVIDIA card over an ATI card any day of the week. I have used ATI almost exclusively and recently switched and have fewer problems and they bring drivers out faster."
Here comes the FUD, I see. Nvidia and AMD both have pretty good drivers. Cards branded as ATI are mostly off the market now, except for the low end.
The difference between the 990FX and 970 chipsets is PCI Express bandwidth. 990FX is good for CrossFire or SLI, but there's no need for it for a single video card.
Why are you still looking to spend $9 on thermal paste, rather than using what comes with the heatsink? The difference that makes will basically be a rounding error. And why a cheap aftermarket heatsink? Is that better than the stock heatsink that AMD ships? It might be, or it might not be, and even if it is better, it isn't much better.
Why are you still looking to pay $90 for 8 GB of system memory? There are plenty of options for that capacity under $50.
But the power supply is the real problem. A Corsair GS series power supply is all right. A Coolmax anything probably isn't. I don't know how bad that particular model is, but surely you don't, either. Coolmax has made enough power supplies that are complete garbage that I'd assume that all Coolmax power supplies are junk until proven otherwise. And don't hold your breath waiting for it to be proven otherwise. It's probably merely mediocre rather than completely awful, as 80 PLUS certification does tend to filter out the really bad ones, but I wouldn't recommend it even if it were cheap. And you're paying more for it than the high end power supply that I linked in my build.
As I said befor on the processor FX8150 in a double blind study felt smoother than a 2700K. Bash the study all you want, it was supervised, there were no AMD banners or tricks. IT IS A BETTER PROCESSOR.
My setup will get him a massive overclock, a faster SSD, and 100$ headroom to upgrade what he sees fit. I'd bump up the Mobo to the Asrock 990fx extreme 4, Or get a higher grade 6870 that has a better cooler.
"990x/FX have 5200 HT link speed vs 4800 for 970. Is good for 8150 which uses HT for core to core communications. I'd even bump up the mobo to the Extreme 4 that has 8+1 power vs 4+1 for the extreme 3"
For a single video card, the extra HyperTransport bandwidth doesn't matter, as you can't use it anyway. I'd be utterly shocked if Zambezi has one core communicate with another by routing the signal through the southbridge rather than merely across the CPU. I'd like to see a reference on that. I can understand wanting more capable power delivery, but does the original poster strike you as a serious overclocker?
"If you read the Thermal grease reviews you would know that Mx2 paste is much better than stock grease. Go read the thermal grease shootout. And that HS with 2 fans in a push/pull will put a H70 to shame. And the HS you linked has abysimal height making it block ram slots."
It's maybe a couple of degrees, if that.
http://www.hardwaresecrets.com/article/Thermal-Compound-Roundup-December-2011/1445/5
And those tests were conducted in a way that will produce larger differences than most people. Note that most of the ones down at the bottom are various household substances for which they said, let's try using this as thermal paste and see what happens.
"Yes but with an unlocked processor and a mobo that supports up to 2100 why not get the faster ram. it does make a difference."
Never mind what the motherboard supports. The memory controller only supports 1866 MHz. And you can get 8 GB of 1866 MHz DDR3 for $60. The real-world performance difference between 1866 MHz memory and 1600 MHz won't often be more than a rounding error anyway.
"That 850W power supply was checked out by Proclockers, 100A, good temps and build quality. And it's Modular."
I'm guessing you're talking about this pathetic excuse for a power supply review.
http://www.pro-clockers.com/powersupplies/701-coolmax-zp-1000-watt-power-supply.html
Look at the hardware they used. They might have pulled 300 W from it, if that. Certainly not even 400 W. That's no way to test a 1000 W power supply. They didn't open it up to inspect the build quality, either.
For comparison, the Rosewill Capstone was tested by Jonny Guru.
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=266
The same platform has been reviewed repeatedly by Hard OCP and Hardware Secrets as well, albeit under other brand names.
I just read it.
The difference between Arctic Silver 5 (widely regarded as one of the best, and tied for 2nd on this web site) and Pink Lipstick was 8C.
I couldn't find anything called MX2 on this list. There was an Arctic Cooling MX-4, which performed pretty mediocrely (and tied with Mayonnaise, the popular condiment, fwiw).
Most every single ~real~ heatsink compound (that is, those marketed for that specific purpose) performed within a 3C band, with none outside of a 5C band, which could very well be argued to be within the margin of error of the test. With one exception: any compound Rosewill brand performed particularly poorly.
Aside from that, I think the only real lesson to take away from this chart is: Use some kind of heatsink compound, it doesn't really matter what though, they are all about the same. If you work on a lot of computers, and want to spring for a tube of the good stuff, or really want that extra edge for very high and tight overclocks, then by all means. But for 99% of people who just want to put together a computer and have it work well, or even with light or modest overclocks, the stock stuff that comes with the heat sink is more than good enough. Heck, even Mayonnaise is good enough...
It's there 5 from the top in excellent performance section
I will agree that SSD is totally unnecessary. It helps, but not so much that it is anywhere approaching worth the additional cost. If money is an object, there is so much else you can spend it on that will help you a great deal more.
But, people are giving you great options for putting your computer together, which is great, but remember that you have to put this together. If you friend can help you, that would be ideal. It isn't terribly hard, but there are something you could do that will destroy your machine before you even turn it on, as a result of static electricity, among other things.
It isn't hard, by any stretch, but you would be better off with some help for your first time out.
"There are two great powers, and they've been fighting since time began. Every advance in human life, every scrap of knowledge and wisdom and decency we have has been torn by one side from the teeth of the other. Every little increase in human freedom has been fought over ferociously between those who want us to know more and be wiser and stronger, and those who want us to obey and be humble and submit."
John Parry, to his son Will; "The Subtle Knife," by Phillip Pullman
"You need to read this. IMO 5 degrees isn't chump change. It's probably less of a difference but applying it to CPU and Video card can help a lot."
It looks like we basically agree that different thermal paste does make a relatively small difference, but disagree over just how big of a difference it is and how much that matters.
51 of the 55 thermal pastes listed there (I'm excluding the various household substances) come in somewhere between 33 and 38 C. That's nearly all of them in a band only 5 C across.
What do you think would happen if you picked one thermal paste and tested it 55 times? Think it would give exactly the same measurement every time? It most certainly would not. There's some variation in room temperature, and in the measured CPU temperature. There's also variation in how the thermal paste is applied, and that can make a difference of another degree or two. The analogous statement of all but a few outliers being in a particular band would probably require a band 3 or 4 C across.
Now yes, 5 C across is bigger than 3 or 4 C. So a better thermal paste probably does tend to make a difference of a couple of degrees or so. Now does a couple of degrees matter? If you're going for a heavy overclock and looking to fry the CPU, then yes, it does. At stock speeds or a moderate overclock, no it doesn't.
"The controller goes higher, I have my PII X6 at 2000 atm.
Not so, look at the benchies.....Cryptograpy gets a boost, 3dmar11, BF3, CivV, MoH. Honestly I agree 2000 vs 1866 or even 1600 is minimal. 1866 would give good compromize and mabye some."
The Zambezi memory controller is only rated at 1866 MHz. Now, you can overclock it, but overclocking a memory controller with a rated speed that is already fairly aggressive (1866 MHz is the highest ever for any CPU, and by a large margin) is a rather different matter from overclocking a memory controller for which the company went curiously conservative on the rating (after Lynnfield was rated at 1600 MHz, do you really believe Sandy Bridge can't do the same?).
And look at your link. They have a confounding problem of not merely using higher clock speeds, but also having tighter memory latencies at the higher clock speeds. So it's entirely possible that even if there were improvements, they would entirely be due to tighter latencies and not the higher clock speed.
And then look at the numbers. Dirt 3 does best at 1866 MHz, and then 1600 MHz, and then 2000 MHz only after those. Crysis 2 does best at 1600 MHz, and performs better there than at 1866 MHz or 2000 MHz. Now, the differences are rounding errors, so it's not like 2000 MHz is outright worse. But it's not better, either, and there's no sense in paying an extra $60 for a rounding error.
I don't see any "Excellent Performance" section. Are you looking at the same link that you provided? Maybe you should post this link because it's certainly not the Hardware Secrets Thermal Compound Round-Up that the rest of us are looking at.
http://www.hardwaresecrets.com/article/Thermal-Compound-Roundup-December-2011/1445/4
This link has 1 page of results, an unsorted chart, and a sorted bar graph.
5 from the top in the unsorted chart is Evercool EX420-TU15, which performed at a mediocre 35C (about average). It ties for 4th place in the ordered chart
5 from the top in the ordered graph is Enermax (Dow Corning TC-5121), which being in 5th place, performed at 33C, which makes it tie for second place.
THe difference between 1 and 8th place are all separated by 1C. Between 8 and 9 is 2C. 9th to 10th is 1C again. Then 10->11 is 19C (no heatsink compound at all), and in last place was chocolate, which was 43C away from 1st place.
This link is over a year old now, but it's pretty applicable.
Tom's takes a Lynnfield i7 and runs 4 different memory configurations on it, from DDR3 800 to DDR3 1600, and each of those speeds with a couple of different latencies.
http://www.tomshardware.com/reviews/core-i7-870-1156,2482.html
Latency has almost as much effect as clock speed.
The jump from DDR3 800 6-6-6 (their baseline) to DDR3 1600 8-8-8 (the highest performance they tested) was 5%. That's doubling the speed.
Their conclusion:
If you can get cheaper RAM for basically the same price, go ahead, it doesn't hurt. But if you are paying so much that you could use that money to upgrade your CPU or GPU, then your wasting your money, because CPU/GPU will yield dramatically better results.
The performance gains are a bit more than a rounding error, but it isn't much. Some programs were more sensitive to it than others, and some systems are more sensitive than others (those that rely on integrated graphics, such as Llano). Some programs showed almost no difference (likely bound by other criteria). SiSoft's various encryption benchmarks showed between 0 and 0.5% difference from the lowest to highest memory speeds tested. The same memory benchmarks showed 178% increase (which is to be expected really, the clock speed has doubled).
For the games they tested, Left 4 Dead showed a decent improvement, about 8%, but it was far and away the largest delta. Interestingly enough, in some tests, higher rated memory performed poorer (DDR1300 10-10-10 being faster by 1.9% over DDR1300 7-7-7 in the GTA4 test, for instance, when they are the same chips just at different BIOS settings) - which just goes to show the margin of error in the testing.
My Overdrive benchmark jumped from 7000 to 7500 points by using ddr3 1750 8-9-8-25 timings over ddr3 1600 7-8-7-24.
It depends on your processor and how much it needs memory bandwidth. My phenom X6 @4ghz was clearly starved for bandwidth at 1600 so jumping up the memory speed yielded very good results. For a quad core jumping up from 1600 isn't a big improvment.
I would assume an 8 core needs the bandwidth even more than a 6 core so using FX8150 with DDR3 1600 is really stupid IMO.
I will get AMD optimized DDR3-2000 with really tight timings when I upgrade to Piledriver next year. I'v heard all the opinions that it makes little difference but when I open a browser with 20 tabs and it just snaps open and loads much faster there isn't much that can sway the real world performance I see.
Every one here has his or her opinion so i'll give you mine too.
SSD drives not neccessary they don't really add that much to the game exp to justify the cost if you ask me and they are really small.
You do not really need a top shelf CPU either Games are not really pushing the envelop of an 8 core system yet and OSs don't either.
I will tell you what I have and why
I started by choseing my chipset. The chipset is what will determine what features you have and can control in your bios settings like overclocking, variable fan speeds, and lots more
Chipset - AMD 790Gx
Socket is also important it will determine which processors you can fit to your MB
GA-MA790GP-DS4H (AM2+)- This is a really solid MB with a huge range of CPUs that fit it. About 3 years ago it was (not sure if they are still available) This is an old MB and its biggest detractor is RAM speeds they are slow DDR2 is old tech. This would be one very good reason to move to a newer MB $150
CPU: I started with a duel core 7750 black edition and later upgraded.(great for overclocking and easy with the chipset features in the bios) think this one ran about $89 back then
NewerCPU $179
vendor
CPU Model
Frequency
L2
Cache
L3
Cache
Core Name
Process
Stepping
Wattage
System
Bus(MT/s)
Since BIOS Version
AMD
Phenom II X6 1100T
3300MHz
512KBx6
6MB
Thuban
45nm
E0
125W
4000
F7I
RAM: Dual CHannel DDR2 1066, the MB can support up to 16gig of RAM but I've never found 4gig sticks in this flavor so I stuck with 8 gig. This old tech for memory and would be a good reason to look at some of the newer MBs and chipsets.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820104073 at $126 for 4x 2 gig sticks
Case: http://www.newegg.com/Product/Product.aspx?Item=N82E16811156063 $79
Vid card: this is where opinions vary massively. Any vid card that can run between 30 and 60fps is all you'll ever need and you can get this for less than $200.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814102948 $164
Power supply: I am not a pro when it comes to these a 750watt for this system will suffice
http://www.newegg.com/Product/Product.aspx?Item=N82E16817139010 $150
HDD: http://www.newegg.com/Product/Product.aspx?Item=N82E16822152291 $120
This is not a top of the line best of everything but it is a solid system I built rought 3 years ago and it will be good for about another 3 years if i have anything to say
I am assuming you will reuse some of your old gear like the CD, mouse, keyboard, monitor, maybe even a fan or two.
This system provided you can get all the parts will run you about $1168
AMD Phenum II x4 3.6Ghz 975 black edition
8 gig Ram
Nvidia GeForce GTX 760
Socket AM2+ has been obsolete for about three years, and does not support modern processors or even modern DDR3 memory. That doesn't mean you need to scrap it right this minute if you already have it, but it does mean that you shouldn't buy it new.
Thuban processors from around April 2010 can go into some socket AM2+ motherboards, but even those seem to be discontinued and disappearing. Zambezi processors are meant for Socket AM3+, and while they can go into some AM3 motherboards, they can't go back to AM2+, as it doesn't have a DDR2 memory controller. Llano processors are Socket FM1, which isn't backward compatible to anything before it. Obviously, Intel processors wouldn't go into an AMD motherboard, either.
I doubt you can get this stuff now as I mention it is a 3 year old sysem and it is what I run atm. I am simply demonstraiting that my 3 year old system is still in the running and will be for at least a few more years before it is totally obsolete.
I am going to build a new system in the near future, I jsut have not finished researching the parts I want to put in it.
AMD Phenum II x4 3.6Ghz 975 black edition
8 gig Ram
Nvidia GeForce GTX 760