A question to those of you with a HD7950 (boost edition)

[Danteska]

Had a Sapphire HD7950 3GB DDR5 Vapor-X OC 950/1250 boost edition for a bit over a year now and I feel like the card is underperforming. This card was first on a 8120 processor and then on a Phenom II X4 955, so I thought it would be just the bottleneck, but I've now tried it on many different PCs, such as my FX6350 or an i5 4690 and even an i7, and the results are always the same. Tried it on four different games and it's giving 20-30 fps less than it should. For example, Tomb Raider with these settings,1080P, 14.4 drivers (tried 14.9, 14.11.2 and 14.12 and 14.4 are the ones giving me the best results; 14.9, 14.11.2 and 14.12 give me 10-15 fps less) and Vsync off should be giving me 90-100 fps on avg on its benchmark, yet I'm barely getting 70.
Here's my 3dmark 11 performance test score:

Could someone else, who got this GPU too, share their avg FPS on other games benchmarks? Or even TR if you got that game. If I have the game you share your score of, I'll try benchmarking my GPU there too, to compare scores. If it's really underpeforming, I'll send it to the RMA service, see if they can fix it or swap it.

 

[Law-II]

Hi

Edit the OP and *include the following:

What *Resolution?
Which *Drivers version?

answering these questions will help other members of the TPU community assist further

Note: when was the last time the dust was cleaned out from the fan/heat sink on the vga card

atb

Law-II

 

[EarthDog]

Its tough to compare Futuremark benchmarks since the CPU plays a pretty big role in the overall score. That said, it certianly looks like its in the ballapark... note you can compare with like hardware at the futuremark website............

Take a look at the Unigine Heaven and Unigine Valley threads here and compare your results there.

 

[Danteska]

Hi

Edit the OP and *include the following:

What *Resolution?
Which *Drivers version?

answering these questions will help other members of the TPU community assist further

My bad. Added.

Note: when was the last time the dust was cleaned out from the fan/heat sink on the vga card

atb

Law-II

Not too long ago, card is practically clean, but what would a lot of dust have to do with bad performance?

Its tough to compare Futuremark benchmarks since the CPU plays a pretty big role in the overall score. That said, it certianly looks like its in the ballapark... note you can compare with like hardware at the futuremark website............

Take a look at the Unigine Heaven and Unigine Valley threads here and compare your results there.

I did take a look there, but all cards there are overclocked, so I can't really compare. Also, that thread doesn't specify whether the 7950 is the old normal version or the boost one with 950/1250 stock values.
Here are my Unigine Heaven 4.0 results tho:

 

[Law-II]

Hi

Not too long ago, card is practically clean, but what would a lot of dust have to do with bad performance?

thermal throttling, GPU has ability to down volt/clock if it gets to hot {threshold is around 72, maybe a little higher} [dust in the fan/heat sink would allow the throttle to activate early and would effect performance]

Note: good quality third party coolers have a direct impact at keeping temps under control

Edit: check the *power settings in ccc increase them to +20 [*if throttling]

atb

Law-II

 

[INSTG8R]

Hi


thermal throttling, GPU has ability to down volt/clock if it gets to hot {threshold is around 72, maybe a little higher} [dust in the fan/heat sink would allow the throttle to activate early and would effect performance]

Note: good quality third party coolers have a direct impact at keeping temps under control

atb

Law-II

Vapor-X is one of the best coolers you can get. I really don't think his cooler is an issue. The threshold is probably closer to 80-90 certainly not 72

 

[Batou1986]

Well if there is a problem mine is worse than yours
FYI R9 280 is a 7950 everything stock and mine is a non boost version

 

[Danteska]

Hi


thermal throttling, GPU has ability to down volt/clock if it gets to hot {threshold is around 72} [dust in the fan/heat sink would allow the throttle to activate early and would effect performance]

Note: good quality third party coolers have a direct impact at keeping temps under control

atb

Law-II

Humm, now that you mention it, here's a picture of what I got:

And most reviews that tested this version of the card did mention its normal full load temperature is around 75-77ºC. Thing is mine's unbelievably cold when at full load: it barely reaches 55-60ºC, 65ºC as max if there's an FPs redundance (like in games like Dead Space, where you get 300-400 FPS and you need to use D3Override to force vsync), but that could of course be due to a good fan speed controlling setup.

I could try showing you guys the volt values when at full load, if you know which ones should be the proper ones, because I don't.

Well if there is a problem mine is worse than yours
FYI R9 280 is a 7950 everything stock and mine is a non boost version

I heard that Windows 8.1 has way better performance than 7, so who knows, maybe yours is fine Also, what are the stock values of that R9 280, are they something like 850/1250?

 

[RadFX]

My bad. Added.

Not too long ago, card is practically clean, but what would a lot of dust have to do with bad performance?

I did take a look there, but all cards there are overclocked, so I can't really compare. Also, that thread doesn't specify whether the 7950 is the old normal version or the boost one with 950/1250 stock values.
Here are my Unigine Heaven 4.0 results tho:

I don't see any problems. Your Temps are good and your scores look fine. I just tested my R9 280 Oc'ed to 1.1ghz gpu and 1.5ghz Ram and got this:

Stock is 933mhz Gpu 1250 Ram @ 1.250v Oc'ed is 1.1ghz Gpu 1.5Ghz Ram @ 1.168v

 

[Batou1986]

Humm, now that you mention it, here's a picture of what I got:



And most reviews that tested this version of the card did mention its normal full load temperature is around 75-77ºC. Thing is mine's unbelievably cold when at full load: it barely reaches 55-60ºC, 65ºC as max if there's an FPs redundance (like in games like Dead Space, where you get 300-400 FPS and you need to use D3Override to force vsync), but that could of course be due to a good fan speed controlling setup.

I could try showing you guys the volt values when at full load, if you know which ones should be the proper ones, because I don't.



I heard that Windows 8.1 has way better performance than 7, so who knows, maybe yours is fine Also, what are the stock values of that R9 280, are they something like 850/1250?

940/1250 is stock i have the same cooler with a slightly different shroud

 

[INSTG8R]

Trust me Batou the Dual-X is nothing like the Vapor-X. Completely different design.

 

[Law-II]

Hi

I could try showing you guys the volt values when at full load, if you know which ones should be the proper ones, because I don't.

Post them here; one of other members of the community will assist with the volts I am sure of that.

Edit: assume have all windows power saving features turned off

nb: sweet looking card

atb

Law-II

 

[Batou1986]

Trust me Batou the Dual-X is nothing like the Vapor-X. Completely different design.

Ok well it doesn't seem to make a difference temp wise maybe I just have good airflow.
>from that heaven run

 

[INSTG8R]

Ok well it doesn't seem to make a difference temp wise maybe I just have good airflow.
>from that haven run

Yeah I got a Dual-X too it's great but the Vapor-X is far superior. My only regret was not waiting long enough for the Vapor-X to come out when I bought mine.

 

[Danteska]

Hi

Post them here; one of other members of the community will assist with the volts I am sure of that.

The screenshot Batou1986 posted was taken from Aida64, right? Looks like that logs min/max/avg values from while open, would that be enough or should I better post the messy .txt file techpowerup gpu-z generates?

Edit:[/B] assume have all windows power saving features turned off

Both my minimum and maximum CPU energy are set to 100%, so CPU is unparked. PCIE energy savings are off. Anything else I should modify?

nb: sweet looking card
atb

Law-II

It's indeed really good looking, but not that good performance wise I think I had higher scores with a 5970 in Unigine, lol. GTX 670 runs all games on my pc so much smoother that I'm about to sell this and buy a used 670.

I don't see any problems. Your Temps are good and your scores look fine. I just tested my R9 280 Oc'ed to 1.1ghz gpu and 1.5ghz Ram and got this:

View attachment 60838 Stock is 933mhz Gpu 1250 Ram @ 1.250v Oc'ed is 1.1ghz Gpu 1.5Ghz Ram @ 1.168v

Mind trying with 14.4 drivers to compare better? Pity neither I got windows 7 nor you got windows 8.1 :/
My temps might look good, but I'm afraid they might even be too low.

 

[Law-II]

Hi

Anything else I should modify?

Just that power setting in CCC; try with +20

Edit: maybe worth reinstalling DirectX

GTX 670 is another fine card [my only beef with NV is streaming video; to me looks a little pixelated in comparison to AMD who use some wicked algorithms to sharpen up the image]

Edit: is the pcie slot running at the correct speed, can check this with GPU-Z

atb

Law-II

 

[Danteska]

Hi


Just that power setting in CCC; try with +20

atb

Law-II

Tried that too when doing the tests, gave me +5 fps on average.

 

[Batou1986]

Hi


Just that power setting in CCC; try with +20

atb

Law-II

this

 

[Danteska]

Edit: maybe worth reinstalling DirectX

How do I uninstall it first? Or does the installer come with an option to delete everything before installing over?

GTX 670 is another fine card [my only beef with NV is streaming video; to me looks a little pixelated in comparison to AMD who use some wicked algorithms to sharpen up the image]

Imho, ATI/AMD cards are fine too, hardware wise, it's just that their software, drivers that is, sucks, drivers and their profiles are far behind in comparison with nvidia.

Edit: is the pcie slot running at the correct speed, can check this with GPU-Z

atb

Law-II

Yeah, a friend of mine got a 6870 and his PCIE slot is bugged or whatever, as it's a PCIE x16 2.0 working @ x8 1.0 even when at full load, but mine is working correctly:

I first also thought it could be a problem due to a 2.0 PCIE slot instead of a 3.0, but then I checked some thorough tests and all showed only 3-5 fps improvement over 2.0, so it's not because of the 2.0.

 

[EarthDog]

What is the problem? Scores look fine to me? If the scores are ok, why are we reinstalling DX?

That 7950 is bottlenecked by 1% on PCIe2 x16 vs 3.0 x16... not an issue.

 

[Danteska]

What is the problem? Scores look fine to me? If the scores are ok, why are we reinstalling DX?

Main problem was (and I use was because looks like it's the GPU or the drivers themselves, that can't give more) the card should be performing same as the GTX 670 or there should be very little difference, 5-10 fps max. I've tried both GPUs on my PC and 670 is performing way better, when both should perform almost equally. For example, while with the GTX 670, I can play The Crew with Ultra graphics (Vsync off and FXAA instead of 4xMSAA because AA in that game is broken) at a stable 60 fps everywhere and the 7950 can barely reach 40-45 fps.

That 7950 is bottlenecked by 1% on PCIe2 x16 vs 3.0 x16... not an issue.

Yeah, I ditched PCIE 3.0 as a problem like week ago, know it's not an issue

 

[Law-II]

Hi

How do I uninstall it first? Or does the installer come with an option to delete everything before installing over?

I understand that this may or *may not be necessary; however DirectX cannot be uninstalled as it is integrated within the OS.
Using steam as an example: DirectX will install with each game that is installed; verifying and updating DirectX files when necessary keeping the API up to date.

Note: GTX GPU die architecture differs considerably and in games coded for this, they perform very well

atb

Law-II

 

[RadFX]

The screenshot Batou1986 posted was taken from Aida64, right? Looks like that logs min/max/avg values from while open, would that be enough or should I better post the messy .txt file techpowerup gpu-z generates?


Both my minimum and maximum CPU energy are set to 100%, so CPU is unparked. PCIE energy savings are off. Anything else I should modify?


It's indeed really good looking, but not that good performance wise I think I had higher scores with a 5970 in Unigine, lol. GTX 670 runs all games on my pc so much smoother that I'm about to sell this and buy a used 670.


Mind trying with 14.4 drivers to compare better? Pity neither I got windows 7 nor you got windows 8.1 :/
My temps might look good, but I'm afraid they might even be too low.

Kind of a pain to switch. I will tell you though your temps look right. Mine with the card overclocked only reached 58-59 in that test. I have an XFX Double Dissipation card btw. Also your GTX 670 is about the same performance as a factory clocked 7970...which would be somewhat faster than the 7950 boost. You'd be better to crossfire or get a next gen card.