Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
GPU/upgrade advice needed
Hi all,
I recently built a new pc here are the specs:
Case: NZXT Phantom 410
Processor: Intel i5-3570k
Motherboard: ASRock Z77 Extreme4
PSU: Seasonic 620W
SSD: Kingston HyperX 120G
HDD: WD black 1TB 7200 RPM
RAM: Team Xtreme Dark Series 2x4GB DDR3 1600
Monitor: ASUS VE247H 23.6"
OS: Windows 8
I now have $600 to spend on upgrades for this pc. Most notable a GPU is missing. So advice on which gpu would be great. I feel like a $600 dollar gpu would be overkill. I was thinking maybe ~400 gpu and a second monitor? What do you think?
Comments
AMD have all next gen consoles so it would be wise to get an AMD GCN card.
I reccomend the MSI 7970 ghz edition.
Or get a 7870 and wait for better cards.
I would have suggested a GTX 680, but if you dont want to spend it all on one item....I would go for maybe a 670. I have one along with my 3770k processor and I can pretty much play any game maxed. I've heard good things about the 650 Ti as well but your CPU should deserve something better IMO. As far as a second monitor.....is it something you would really take advantage of? If so, then go for it! What kind of sound yout got coming out of your pc? Do you have a good quality sound card and speakers. To me sound is just as important as any other component.
I got that exact Dell display for my birthday last year: I can highly recommend it.
http://www.speedtest.net/result/3845509852.png
There is also a newer version of that card that is slightly more expensive ($20) but has better OC capacity, is consistantly cooler and shorter.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814202026&Tpk=100352-3L
8ms! Are you serious? That is way too slow to game with. If you PvP that monitor will get you killed.
Myths #2 & #3
Response Times: How Fast Is Fast Enough?
It's basically a 7930 in disguise
A capable gaming system will typically have display latency in the ballpark of 40 or 50 ms--meaning, at any given moment in time, the monitor is displaying the state of the game world about that long ago.
Bringing that down a few ms by using a TN monitor would be nice from a display latency perspective, but it will come at the expense of image quality. It's a matter of priorities, and an extra few ms won't often be the difference between winning and losing.
If you want to go all out to bring display latency down, then a 120 Hz monitor is the way to do it. And then you can turn graphical settings down far enough to actually render many games at 120 Hz or better, too. That can get your display latency down to around 30 ms or so. Those don't tend to have all that good of image quality, though.
Considering that a Radeon HD 7970 GHz Edition is both cheaper and faster than a GeForce GTX 680, I wouldn't get the GTX 680. A GeForce GTX 660 or GTX 670 may be worth considering if you want an Nvidia card.
If you want to just get a video card and nothing else, the a Radeon HD 7970 GHz Edition is the fastest that fits the budget. A second monitor is a worthy purchase if you only have one, though.
You mentioned 2 monitors so may want a little more horsepower if that's what you decide. I see a number of AMD responses here so I'll suggest a few from the Nividia camp:
If you are going to run one monitor at say 1920 x 1080 or 1920 x 1200, you would be fine with say a GFX 650 TI or a 660 GFX or 660 TI. The 670 and 680 cards are wonderful but expensive. My personal opinion is something near to a 660 TI is about the best bang for the buck out there.
Everyone has a preference with GPUs. Ultimately it depends on why you plan to run for games. Personally, I play MMOs primarily and RPGs (Skyrim currently) I'm running 2 660 GFX cards (Overclocked) in SLI because I found a great deal on them - couldn't be happier...
With that said, you are usually better off buying 1 card within your price range vs. running a couple but again - it depends.
I'm buildiing a machine (based on an i3 Ivy Bridge) for my wife next week and am going to roll with the newer 650 TI card. It will get the job done for her and I'll save a little money.
If you are a gamer, the GPU is an important decision, almost more so that the CPU at this point given the performance of the last couple generations of Intel processors, Most of us would probably be fine with 2 physical cores (Like an i3) but would be quickly bottlenecked with a crapp GPU. I'm still running a 1st gen i7 950 (Bloomfield) in my main machine. While I always want the latest and greates in my gaming PC, I just can;t make the case to upgrade the CPU but GPUs? That's about a 2 year cycle for me consistently.
Seaspite
Playing ESO on my X-Box
The only reason to consider two cards in CrossFire or SLI is if one high end card isn't good enough for you. I could understand a CrossFire or SLI setup as an alternative to a single GeForce GTX Titan, just because Nvidia priced it at $1000, but not really a CrossFire or SLI setup with anything slower than a GeForce GTX 670 or Radeon HD 7970. One faster card will perform well far more reliably than two slower cards in CrossFire or SLI.
Getting an Ivy Bridge Core i3 CPU together with a GeForce GTX 650 Ti is a big mistake. If she doesn't need much in the way of graphical performance, then save some money and get an A10-5800K. If it is a gaming rig, then go ahead and get the GTX 650 Ti if you find a good deal on it (meaning, substantially cheaper than a Radeon HD 7770, and not meaningfully more expensive than a GDDR5 version of a 7750), but don't get the Core i3.
For the same price as an Ivy Bridge Core i3 that is verging on obsolesense for gaming purposes even today, you could get a six-core AMD FX-6300 that will be good for eyars to come. If you're willing to accept a mediocre processor, then you at least want to pick one with a much lower price tag than Intel's Core i3s.
130303
cpu
i7 3770 4c/8thread $220-280, (3dmark 8260) 3.4GHz/150W (?77W) lga1155 (25.6 GB/s max mem bwidth, graphics 650MHz, hyperthreading);
i7 3770K $380-320, (3d mark 8470) has unlocked multiplier (~35[oc 4.5GHz] 3.5GHz@100W/wo graphics), idles @ 1.6GHz/44W, oc ~4.7GHz air cooling, 23 Apr,2012, 3rd gen, 3 yr war, 77W/67*C, 22nm, ddr3, integrated hd 4000 graphics @ 650MHz, lga1155;
~i7 3820 4c/8th $285, feb'12 (3dmark 10530, oc @4.75GHz/251W) lga 2011, 130-158W, 3.6GHz, 10MB L3, 4ch ddr3 1600, no igpu (therefor no quick sync), slower L3 cache, no oc K series (base clock can be uped to 125 [@4.3 to 5.3GHz {air 37x 125MHz 4.6GHz, vcore 1.45v @80*C}]vs 100MHz), highest multiplier 43x, 40 lane pcie, 32nm;
i7 2600s 4c/8 $306, 2.8GHz, lga 1155, (int grap, 8MB L3) 65W, Jan '11, ddr3 1333 dual Ch, pcie 2,
i7 3960x 6c $990, 3.3GHz, lga2011, 2nd gen sbE, 32nm,
i7 3930k 6c/12th $570, (3d mark 10560) lga2011, 130W, 3.2GHz, 12MB L3;
980x $660, 4c, 3.3GHz, lga 1366, 12M cache, 6.4GT/s, 130W, 32nm
gpu
-gf gtx 660 $200 aug'12(192b *3GB* ddr5 $320), pciE3x16, 980MHz, 2.1Tflop;
-gtx 660 TI $290, aug'12(192b, dp1.2/2dviD/hdmi) 2460 Gflop SP;
680 mar'12 $450(poor availability) (256b, 2&4GB gddr5) 2dulLkdvi/hdmi/dp1.2, 195W tdp, 3Tflop (129Gflop Dbl Pre);
gtx 670 overpriced$330 july'12 (256b, 2&4GB gddr5) dp1.2/2dviD/hdmi, 3yr war, 6.8-9.5", 6&8pin, 295-145W/71-30*, 2.46Tflop
*hd7970 $380 dec'11 (384b 3GB gddr5) 28nm 925MHz clock/1.4GHz memory, > perf 670, 350-89W/60-31*C, 2mini dp1.4a, dual link dvi, hdmi, 10.8x4.4x1.5", <4095x2160res, 3yr war, 6&8 pin pwr, 3.7Tflop SP (947Gflop DP);
7950 $300 jan'12 (384b 3GB gddr5) 28nm 900MHz/ (low multi high res display output?), 2.9Tflop (717Gflop DP)
memory low profile samsung 30nm ddr3 1600 (16GB) $30/4GB...
...
$960
cheapo 27" monitor Acer s271hl $250 +$50 3yr best buy in house war;
i7 3820 $285;
ATI hd 7970 $380; (the 7970 allows high res multi monitor, vs cheaper 7950)
gtx gpu are subpar in dbl precision gflop vs the hd
The only thing i have to say to that post:
Double precision isn't used in gaming, it's mainly for HPC & commerical applications (FireGL/Quadro drivers, CUDA/OpenCL, etc). OpenGL/DIrectX don't require double precision, and many GPUs lack dedicated double precision hardware (typically only present in upper-tier GPU dies that get mirrored in the professional card lineup).
Double precision doesn't even factor into a consumer gaming GPU decision.
Actually, the OpenGL 4.0 and later specification does require double precision support. (I suspect that the reasoning for this is that if the cards were going to support it anyway for reasons of OpenCL or CUDA, they figured they might as well expose it in OpenGL just in case someone wants to use it.) It does not, however, say how fast a video card has to run it. Most consumer cards run double precision computations at 1/24 to 1/16 of the speed of single precision computations, presumably by taking several passes to do the computations.
I wouldn't be surprised if a relative handful of games eventually make light use of double precision computations on the GPU. That wouldn't really be any more far-fetched than my current use of uints for a quasirandom number generator (for particle effects, I want lots of particles with some particular distribution, but don't particularly care where each particle goes, so long as it is consistent), and we have enough flexibility now that you can't really predict what someone will find useful.
But I agree with your conclusion that it doesn't factor into a consumer gaming GPU decision. If games use double-precision computations at all, it will probably be light enough use of it that even cards that are "bad" at it can run the game just fine. And cards that don't support double precision computations at all are old enough that they're not what you'd want to buy today--not necessarily even used.
Good to see your still haunting the tech forums confusing the hell out of us Quizz lol