Sign in with
Sign up | Sign in
Your question
Solved

GTX 580 in an older PC

Last response: in Systems
Share
August 2, 2011 10:29:59 AM

I want to throw a GTX 580 into a PC I built a few years ago. I would like to wait until next year to build an entirely new PC, but 2x 8800 GTS's aren't going to cut it in the meantime.

I think the only limiting factor is my trusty 700W Rocketfish PSU (RF700WPS with 4 x 18A rails)
http://www.hardwaresecrets.com/article/556/1
Power specs in this image

Current Mobo: http://www.newegg.com/Product/Product.aspx?Item=N82E168...
and CPU (OC'd to 3.0GHz): http://www.newegg.com/Product/Product.aspx?Item=N82E168...

Can it handle this 580? http://www.newegg.com/Product/Product.aspx?Item=N82E168...


More about : gtx 580 older

a b B Homebuilt system
August 2, 2011 12:03:52 PM

At 3.0ghz you would have quite a large bottleneck. Maybe overclock more?

What resolution"?
m
0
l
August 2, 2011 12:11:32 PM

Im very fussy in my choice of PSU after using a cheaper brand a few years back. Most people will tell you to stick to Corsair / Antec / Be Quiet etc.... I havent heard of Rocketfish (maybe not avaialble in the UK)

m
0
l
Related resources
August 2, 2011 12:44:04 PM

Yes, you might be slightly bottlenecked by your 3GHz CPU but an updated GPU can still give you a good performance boost.
Also, by running at higher resolutions and settings the bottleneck is moved back from the CPU to the GPU.
If you are able, try and push the CPU to 3.2-3.6GHz to alleviate any potential bottlenecks and, if not, do not give it to much worry.

While your PSU does look solid enough to power a GTX 580, it unfortunately does not have the proper power connectors.
A GTX 580 requires 1x6-pin and 1x8-pin PCI-E power plug, your PSU only has 2x6-pin plugs.

I would recommend instead getting either a GTX 570 or 6950 2GB (preferably unlocked to a 6970 ;)  ).
Either will give you massively better performance/price ratio and both only require 2x6-pin PCI-E power plugs.
m
0
l
August 2, 2011 12:46:21 PM

How large of a bottleneck are we talking about? I'm using a stock intel cooler, so I would have to buy an actual cooler to OC it more.
I run 1920x1080
Rocketfish is supposedly just rebranded something else. The PSU has lasted me a while without any problems. My next PSU purchase will be from a reliable name though.
m
0
l
a b B Homebuilt system
August 2, 2011 12:54:58 PM

n00biee said:
How large of a bottleneck are we talking about? I'm using a stock intel cooler, so I would have to buy an actual cooler to OC it more.
I run 1920x1080
Rocketfish is supposedly just rebranded something else. The PSU has lasted me a while without any problems. My next PSU purchase will be from a reliable name though.


i'd say that you'd be losing about 14FPS or so, the Q6600@3Ghz is still quite capable, may i ask when your going for such a high-end card?
m
0
l
a b B Homebuilt system
August 2, 2011 12:56:39 PM

It's a Huntkey 650w rebranded as Rocketfish 700w. Huntkey is known for being some of the worst psu's on the planet, although that particular model is one of the better ones.
m
0
l
August 2, 2011 1:14:55 PM

ps3hacker12 said:
i'd say that you'd be losing about 14FPS or so

Wow
Way to go pulling numbers out of.. the sky.. :pfff: 

At 1920x1080 your CPU bottleneck, if you have one at all, will be quite minor especially if you have all the settings turned up.
CPU bottlenecks are mostly manifested at very low resolutions with the settings turned down or in low complexity games that are already hitting extremely high FPS.
A 3GHz Core 2 Quad is still quite a powerhouse and I personally would not think twice about keeping it in a high end gaming rig.
An aftermarket cooler and higher overclock would be nice but is hardly necessary.

Yes, your Rocketfish 700W is actually a re-branded Huntkey 650W.
Not really top notch but, as shown in Hardware Secrets Review, it is a capable enough performer (if you do not mind the low efficiency).
It has just enough 12V amperage (42W combine) to meet nVidias recommendations for a GTX 580.
The real issue with it is the lack of a 8-pin PCI-E power plug, which would be required for a GTX 580.
m
0
l
a b B Homebuilt system
August 2, 2011 1:19:03 PM

outlw6669 said:
Wow
Way to go pulling numbers out of.. the sky.. :pfff: 

At 1920x1080 your CPU bottleneck, if you have one at all, will be quite minor especially if you have all the settings turned up.
CPU bottlenecks are mostly manifested at very low resolutions with the settings turned down or in low complexity games that are already hitting extremely high FPS.
A 3GHz Core 2 Quad is still quite a powerhouse and I personally would not think twice about keeping it in a high end gaming rig.
An aftermarket cooler and higher overclock would be nice but is hardly necessary.

Yes, your Rocketfish 700W is actually a re-branded Huntkey 650W.
Not really top notch but, as shown in Hardware Secrets Review, it is a capable enough performer (if you do not mind the low efficiency).
It has just enough 12V amperage (42W combine) to meet nVidias recommendations for a GTX 580.
The real issue with it is the lack of a 8-pin PCI-E power plug, which would be required for a GTX 580.


actually i based it on this :D  :
http://www.anandtech.com/bench/Product/49?vs=288

so yeah about 14FPS or so compared to newer generation processors.
m
0
l
August 2, 2011 2:17:29 PM

Well, it is always good to have a bit of backup for your claims [:jaydeejohn:5]

However, if you dig a little into the actual gaming benchmarks, it does not really point towards a noticeable bottleneck for the OP.
So, just because it is a slowish day at work, I will break it down a little and try to show why the OP should not need to worry :D 

Gaming Benchmarks shown at Anandtech, Q9650 vs i5 2500K:

Fallout 3
1680x1050
Medium Settings

Q9650: 90.3 FPS
i5 2500K: 86 FPS

Difference of 4.3 FPS, Q9650 is above 60 FPS.

Left 4 Dead
1680x1050
Max Settings (no AA/AF)

Q9650: 125.2 FPS
i5 2500K: 142.7 FPS

Difference of 17.5 FPS, Q9650 is above 60 FPS.

Far Cry 2
1680x1050
Medium Settings

Q9650: 61.9 FPS
i5 2500K: 78.9 FPS

Difference of 17 FPS, Q9650 is above 60 FPS.

Crysis Warhead
1680x1050
Medium Settings

Q9650: 83.4 FPS
i5 2500K: 91.6 FPS

Difference of 8.2 FPS, Q9650 is above 60 FPS.

So, looking at only these four games that are tested at a medium resolution and with medium settings, there is an average difference of 11.75 FPS between the Q9650 and an i5 2500K.
In ALL of the tests the Q9650 was able to stay above the critical 60 FPS mark (the refresh rate, and ergo frame rate, of the majority of LCD displays).
Also keep in mind, the games that show the largest differences are not very resource demanding at all (which is why they are chosen for a CPU benchmark).
Light games such as these generally demand very little from the GPU and performance swings heavily depending on the CPU.

Now, when you raise the resolution to 1920x1080 it is only going to effect the graphics subsystem.
That is to say, assuming the GPU is powerful enough to keep the bottleneck on the CPU, there would be NO decrease in frame rates by raising the resolution.

By the same accord, raising the in game settings predominately effects graphical performance with a relatively minor additional load on the CPU.
Again assuming a GPU that is powerful enough to keep the bottleneck on the CPU, there would be either no decrease in performance or a very minor decrease in performance by raising the settings.

Unfortunately for us, even the most powerful GPUs falter at higher resolutions and settings in newer games.
When this happens the bottleneck will move away from the CPU, effectively giving his overclocked Q6600 the same performance as a more current and powerful CPU.

Of course there will probably be some exceptions to this but, overall, a 3GHz Core 2 Quad will perform admirably with a high end GPU at high resolution and settings.
m
0
l
August 2, 2011 3:10:31 PM

I also completely agree with outlw6669. Your Q6600 at 3.0GHz is not really a bottleneck in almost every game at high-settings and 1080p resolution.
m
0
l
a b B Homebuilt system
August 2, 2011 3:39:33 PM

I disagree, it's definitely going to be a bottleneck. I would use the $600 to get a motherboard, cpu, ram and ssd. Then worry about a GTX580 when the price drops. A GTX580 is way too high end for that setup.

Those benchmarks have no relation to this conversation because A)wrong resolution B)Q9650 is faster clock for clock than Q6600 C)You have no idea what video card they are using here D)those benchmarks are all on Medium settings E)AA/AF is probably reduced or off in those benchmarks.

Now who here has a GTX580 @ 1080p and runs medium settings with no AA/AF? Anyone?

m
0
l
August 2, 2011 4:03:12 PM

IMO the benchmark there is for CPU benchmarking thats why they are using medium-settings; and OP wants to know about CPU bottleneck with games.

Isn't running games with high-details and high-resolution is GPU intensive (not CPU) ?

My system when running Crysis at 1080p with everything MAX with 8xAA (and now 4XAA) is giving 35+FPS with Radeon 6970 _and_ my Phenom 1090T @ 3.6GHz is not even utilized fully (I haven't monitored its utilization but my CPU temperature is maximum touching the 47C and in 100% utilization is going upto 59C during Prime95; thats why I am assuming that its not even utilized 60%)
m
0
l
a b B Homebuilt system
August 2, 2011 4:08:51 PM

6970 is slower than GTX580, and your cpu is way faster than his. Not seeing the relevance. Not to mention your guessing what your utilization is, you have no facts.

http://www.techspot.com/review/379-crysis-2-performance...

Yes higher res means more load on graphics card, but that card cant throw frames like it wants if the cpu can't keep up.

There's one easy way to test. Run a benchmark. Overclock your cpu 200mhz and run the benchmark again. If your fps went up after the overclock, you were cpu bottlenecked.

Besides, doesn't your X6 "turbo" when not using all cores?

m
0
l
August 2, 2011 4:45:14 PM

@GeekApproved:

You are right, also comparing GTX580 and my 6970 is not correct.

For my CPU utilization: Turbo is disabled as I already overclocked it to 3.6GHz and CPU-NB @ 2.4GHZ.
m
0
l
August 2, 2011 9:41:38 PM

geekapproved said:
Those benchmarks have no relation to this conversation because
A)wrong resolution
B)Q9650 is faster clock for clock than Q6600
C)You have no idea what video card they are using here
D)those benchmarks are all on Medium settings
E)AA/AF is probably reduced or off in those benchmarks.

F [sorta])Now who here has a GTX580 @ 1080p and runs medium settings with no AA/AF? Anyone?

A) As addressed above, raising the resolution will only serve to bring any bottleneck towards the CPU.

B) Yes, the Wolfdale based Q9650 is, on average, 6% faster clock for clock than the Conroe based Q6600. The Conroe arch is roughly comparable to the current Phenom II arch on a clock for clock basis.

C) Anand's CPU bench uses a eVGA GeForce GTX 280 universally. This offers performance that is roughly comparable to a modern GTX 460 768MB (and incidentally quite similarly to the OPs 8800GTS SLI), or Around 60% of the GTX 580s Performance.

D/E) Again, as addressed above, maxing out the settings (including AA and AF) will put massively more strain on the graphics card, firmly pulling the bottleneck away from the CPU.

F) Exactly my point. When you have a high end GPU you will most defiantly be turning the settings up as far as possible. As such, any potential performance impact from the CPU will be minimized.


But, as they say, the pudding is in the benchmarks (or something like that :p ).

So, Here is a nice review I dug up on the matter.
It investigates CPU scaling with several GPU combinations from high to mid resolutions.

Be sure to pay particular attention to the 1920x1200, Phenom II x4 3.2GHz, GTX 480 tests as they would closely match the OPs system if he where to get a GTX 570 (as already covered, he could not run a GTX 580 on his PSU).
In the vast majority of games, the 3.2GHz Phenom II performs on a similar level to a 3.8GHz i7.
It would take very little tweaking for the OP to hit 3.2GHz and he could probably even hit that with his stock cooler (but should none the less get an upgraded cooler).

So, with benchmarks to backup common knowledge, I stand by my assertions that the OP would be well served with an upgraded GPU even with a dated CPU.
m
0
l
August 2, 2011 9:54:52 PM

geekapproved said:
6970 is slower than GTX580, and your cpu is way faster than his. Not seeing the relevance. Not to mention your guessing what your utilization is, you have no facts.

http://www.techspot.com/review/379-crysis-2-performance...


Also, please do not cherry pick benchmarks then hold them up as the shining truth.
Overall, the GTX 580 performs 10-15% better than the 6970 2GB, not the 24-35% your link shows :pfff: 


m
0
l
a b B Homebuilt system
August 2, 2011 10:10:38 PM

I put a GTX470 in a Q6600 and got very good gain over a pair of 8800gt superclocked. I say do it!
I am using an apeva beast 680Watt. Runs great and the system only pulls like 500watts from the outlet using my kill-a-watt.
m
0
l
August 2, 2011 11:03:48 PM

So I could buy a new PSU and a 580 to get by until I build a new computer next year?
m
0
l

Best solution

August 3, 2011 8:17:33 AM

n00biee said:
So I could buy a new PSU and a 580 to get by until I build a new computer next year?

Yes, you could do that.

I think however you would be better served by a somewhat less expensive card now and an updated one next year when you rebuild the entire system (next gen cards should be out end of this year or beginning of next).
The cards I would again recommend are a GTX 570 or 6950 2GB.

A GTX 570 still has around 90% of the GTX 580s performance but can be found for only $290 as compared to the GTX 580s $440 (current prices from newegg).

A 6950 2GB offers around 80% of the GTX 580s performance and can be found for $245 on newegg.
What is interesting about the 6950 2GB is that, in most cases, it can be Unlocked into a 6970 which would then perform nearly identically to the GTX 570.

Both of these cards would work just fine with your current PSU.

The choice is of course yours.
However I strongly believe that you would be better served with a slightly slower GPU now while saving your current PSU budget and the $195-150 GPU price difference towards your future build.
Share
August 3, 2011 8:26:32 AM

I really appreciate all the help.
m
0
l
August 3, 2011 8:27:11 AM

Best answer selected by n00biee.
m
0
l
August 3, 2011 8:33:12 AM

Not a problem.
Enjoy your updated system!
m
0
l
a b B Homebuilt system
August 3, 2011 6:26:59 PM

Cherrypick bencharks? LOL

I picked the only one that showed 1080P at extreme settings which is relevant to our conversation because anyone running a $600 video card is not going to running medium or high settings.

I still disagree, and the OP can educate us once he gets his GTX580. All he has to do is run a benchmark at 3ghz, then overclock the cpu to 3.2ghz and run the benchmark again. If the average fps goes up, then the cpu was holding back the gpu.

I would bet money there is a difference, but hey, either way he's going to enjoy his $600 card, although I think $600 would be much more wisely spent on a 2500K, mobo, ddr3, and a ssd, and he would still probably have enough left over for a new psu. Then with a little overclocking, he would be set for ANY gpu or dual gpu upgrade.
m
0
l
!