PDA

View Full Version : Just installed EVGA 7900GS


Daneel Olivaw
09-20-06, 07:07 PM
Hi guys/girl,

I just installed my new EVGA 7900GS, the one without HDCP and with the reference cooler. I will be posting some feedback here soon, mostly comparing with my previous card, the great MSI 7600GT.

MSI 7600GT:
http://www.msi-computer.ca/images.php?img=1tHK1dTWnt3W2NLkxuPgk9%2FV0JK8wJG%2 FnqOYnJ2cvJzZ09Y%3D&w=300&h=243&page=popup

EVGA 7900GS
http://www.evga.com/products/images/products/overlayImage.asp?PRODID=256-P2-N620-AR&GS=0&SLI=1&width=242

DataMatrix
09-20-06, 07:15 PM
On a sempron... bottleneck :thumbdwn:

J-Mag
09-20-06, 07:38 PM
On a sempron... bottleneck :thumbdwn:

You CAN'T generalize like that... Well you just did, but your statement retarded, because it is obviously dependent upon the application and its settings. However, thats besides the point.

For most (maybe 98%) of the games out there 2.5ghz is plenty, the loss of 256k cache vs standard Athlon 64's is negligible. Anyway most people play in GPU limited scenario right?

Daneel Olivaw
09-20-06, 07:47 PM
On a sempron... bottleneck :thumbdwn:
At 2.5GHz??? I think not! If you look here: http://www.nvnews.net/vbulletin/showthread.php?t=72826 you will see that I score roughly 300, which is more than overclocked AMD X2 3800+. The only code my Sempron64 is slower at running than any other PC is multithreaded code, and my Sempron64 is my gaming box, so no problems there.

In any event...

3DMark2003
http://www.nvnews.net/vbulletin/attachment.php?attachmentid=20442&stc=1&d=1158795493
Looking at the screen shot, it seems that the 7600GT core at 620MHz (OC'ed) is roughly equal in shader power to the 7600GS core at 450 (stock), which makes sense to a degree as one has 12 pipelines (12 ROPs), and the other 20 (though only 16ROPs).

Memory bandwidth is much higher on the 7900, and that seems to be the only advantage currently.

I'll keep running tests, and overclock in the end...

Daneel Olivaw
09-20-06, 08:06 PM
Again similar scenario in 3DMark2005, since the whole of the tests are pretty much shader limited we get a near tie. Note that the 7600GT is overclocked to close of its limit, while the 7900GS is at stock.
http://www.nvnews.net/vbulletin/attachment.php?attachmentid=20443&stc=1&d=1158796692

Daneel Olivaw
09-20-06, 08:24 PM
3DMark2006
In what I feel is a surprising turn of events, the 7900GS pulls ahead of the 7600GT in SM3.0 tests by what could be argued a clear margin.

However, again, both nearly score the same total.
http://www.nvnews.net/vbulletin/attachment.php?attachmentid=20444&stc=1&d=1158797780

On to overclocking the 7900GS!

Daneel Olivaw
09-20-06, 08:37 PM
With the 7900GS at 493/1500 (up from 450/1320), in 3DM2003 I get a healthy 16425 3dmarks (up from 15295).
Note: Rivatuner 2.0 RC16 was not able to unlock locked pipes (though I wasn't really expecting it to be able to).
Note2: The 7900GS has a 'geometric delta clock' of 20MHz, I will try to remove it and see if it helps overclocking.
Note3: Also interesting is that the fan never speeds up very much, <30% in bios. Temps <50C.

Daneel Olivaw
09-20-06, 10:12 PM
It seems like the clocks I will settle on are 520/1500, which is not too shabby.

I'll play some Oblivion at those clocks and see if the purchase was worth it over the 7600GT, I don't believe it was a big step forward, but considering I'll be selling off the 7600GT soon, it's definitely not a bad upgrade all in all.

I'll post final 3DMarks once I'm done with stability testing.

stevemedes
09-20-06, 10:38 PM
I have the same card, very satisfied.
whats your 3md05 score?

Daneel Olivaw
09-20-06, 10:55 PM
I have the same card, very satisfied.
whats your 3md05 score?
7969 on 3DMark05, High Quality and Trilinear specified in driver. I'm not hitting clocks as high as you, I get artifacts (few white dots in 3dMark03 Ragtroll test) when I go over 1500MHz on the ram.

I've settled on 520/1500. I'm quite happy, though it shows my 7600GT in a good light (The 7600GT did cost me more though 2 months ago than the 7900GS today...).

Oblivion feels smoother on the 7900GS than the 7600GT, but not by much, and the same settings were used. I had to back down my memory clocks again, I'm now at 520/1440.

DataMatrix
09-21-06, 05:20 PM
At 2.5GHz??? I think not! If you look here: http://www.nvnews.net/vbulletin/showthread.php?t=72826 you will see that I score roughly 300, which is more than overclocked AMD X2 3800+. The only code my Sempron64 is slower at running than any other PC is multithreaded code, and my Sempron64 is my gaming box, so no problems there.

In any event...

3DMark2003
http://www.nvnews.net/vbulletin/attachment.php?attachmentid=20442&stc=1&d=1158795493
Looking at the screen shot, it seems that the 7600GT core at 620MHz (OC'ed) is roughly equal in shader power to the 7600GS core at 450 (stock), which makes sense to a degree as one has 12 pipelines (12 ROPs), and the other 20 (though only 16ROPs).

Memory bandwidth is much higher on the 7900, and that seems to be the only advantage currently.

I'll keep running tests, and overclock in the end...

That CPUMark99 sucks balls, you can't use a program from 1999 to measure performance from 2006... Take 3DMark2001 for example, that was meant for graphics cards now its CPU based.

Daneel Olivaw
09-21-06, 10:37 PM
That CPUMark99 sucks balls, you can't use a program from 1999 to measure performance from 2006... Take 3DMark2001 for example, that was meant for graphics cards now its CPU based.How's Sisoft Sandra 2005 telling me I have the same integer and floating point CPU performance as an FX-55 then? Listen I don't want to debate that my 60$ CPU is getting me the same framerates as your 200$ CPU, so lets drop it. Congrats on your X2. BTW, why the h* do you have a CPUz screenshot of your 2T ram in your sig? And 3dmark01 is not CPU based, its CPU-limited. Oh and if CPUMark99 sucks balls so bad, why does it show Conroes owning X2s? CPUMark99 evaluates single thread performance, which is a good test for all current games. Oh and btw, in that other forum, you put your foot in your mouth explaining 64bit CPUs, that was ridiculous. Want me to continue? :lol:

back to topic:

Any1 else have an EVGA 7900GS? I'm curious about what your thoughts are on the card. For the price, I'm quite satisfied, I know stevemedes is too. However, I feel there wasn't really much room between the 7600GT and 7900GT in terms of performance to justify this new model. I believe the 7900GT will disappear from the market soon. I think it was announced anyways.

jAkUp
09-21-06, 11:55 PM
Depending on the game and resolutions he plays at, there is definetely some room to gain over a 7600GT.

The 7900GS is a great card for the money. Its basically like a 7800GT. Which also was a great card :D

DataMatrix
09-22-06, 01:03 PM
How's Sisoft Sandra 2005 telling me I have the same integer and floating point CPU performance as an FX-55 then? Listen I don't want to debate that my 60$ CPU is getting me the same framerates as your 200$ CPU, so lets drop it. Congrats on your X2. BTW, why the h* do you have a CPUz screenshot of your 2T ram in your sig? And 3dmark01 is not CPU based, its CPU-limited. Oh and if CPUMark99 sucks balls so bad, why does it show Conroes owning X2s? CPUMark99 evaluates single thread performance, which is a good test for all current games. Oh and btw, in that other forum, you put your foot in your mouth explaining 64bit CPUs, that was ridiculous. Want me to continue? :lol:

I get better performance than an FX-55, you won't get better framerates as me, I just rebenchmarked and got 9300 in 3Dmark05, why do I have a screenshot of my 2T ram? same reason you have a sempron in your sig, besides 2T/1T has barely any impact on framerates, i'd rather have 2T and lower timings of 2.5-3-3-7 (especially for budget ram), 3Dmark01 IS CPU based since it lost its 3D graphics card benchmarking capabilities a long time ago, CPUMark99 sucks so bad because it uses old instruction sets, which I'm sure the C2D excels at, my post about explaining 64bit was true in a way since 64bits of data is transferred rather than 32, and I also said correct me if i'm wrong.

So go back to bed :)

Daneel Olivaw
09-22-06, 10:40 PM
[...]
So you hate me?
and I hate you!
You know what, you know what?
It's all in the family...
[...]

Daneel Olivaw
09-23-06, 09:41 AM
My point was that my overclocked Sempron is equal to an FX-55, regardless of the cache size, in games. If I had said I had an FX-55, you would never have said that my CPU would be a bottleneck.

And check your goddam definitions, 3Dmark01 is not CPU-BASED, to say that would mean that 3Dmark01 was developped intentionally to test CPU, that is just plain wrong. It is CPU-LIMITED, or CPU-BOUND, now and wasn't 5 years ago!

If you're proud of your cas 2.5 budget RAM, then you should appreciate a 2.5GHz Sempron.

On 64bit computing:

At this point, I should make a quick note of a fact that I'll refer to again in the article's conclusion: increased performance was not mentioned above as a straightforward, across-the-board benefit of increased dynamic range. As I stated previously, 64-bit integer code runs slowly on a 32-bit machine, due to the fact that the 64-bit computations have to be split apart and processed as two separate 32-bit computations. So you could say that there's a performance penalty for running 64-bit integer code on a 32-bit machine; this penalty is absent when running the same code on a 64-bit machine, since the computation doesn't have to be split in two. The take-home point here is that only applications that require and use 64-bit integers will see a performance increase on 64-bit hardware that is due solely to a 64-bit processor's wider registers and increased dynamic range. So there's no magical performance boost inherent in the move from 32 bits to 64 bits, as people are often led to think by journalists who write things like, "64-bit computers can processes twice as much data per clock cycle as their 32-bit counterparts." Technically, this is true in a very restricted sense, but it would be better to say the following: "64-bit computers can process numbers that are 4.3 billion times as large as those processed by their 32-bit counterparts." It sounds a lot less sexy because it is, but at least no one is misled into thinking that 64-bitness makes a computer somehow twice as fast.

And it should be noted that gaming rarely computes 64bit datatypes (at least on the CPU). And that a 64bit register can still only hold one value, whether it be a value fitting in the 32bit range or 64bit range.

You now stand corrected on 64bit computing. I wanted to check my info, as I had studied only 16bit and 32bit computing at the University. But the differences between 16bit and 32bit are also true between 32 & 64.

References for 64 bit computing:
http://www.chip-architect.com/news/2003_09_21_Detailed_Architecture_of_AMDs_64bit_Cor e.html
http://arstechnica.com/cpu/03q1/x86-64/x86-64-1.html

About CPUMark99 sucking because of old instructions sets, do you know what percentage of modern day games/apps uses instructions above the set of SSE? Nearly 0.

Integer additions and compares (mults too) are whats going on in a game (or any other app) 99.99% of the time, thats what CPUMark99 does. And if the Core 2 is so much stronger at that than your X2, then it'll be so much stronger in every other apps too.

Go back to school kid.

Daneel Olivaw
09-23-06, 09:55 AM
http://www.nvnews.net/vbulletin/attachment.php?attachmentid=20566&stc=1&d=1159226845

Daneel Olivaw
09-25-06, 07:34 PM
Holy ****! With the 1.4V voltmod, I'm getting a strong 620MHz on the core!!!!

J-Mag
09-25-06, 07:47 PM
LOL you got 5 posts in a row and then 3 posts in a row... I don't think i have ever seen that before.

crainger
09-25-06, 07:52 PM
LOL you got 5 posts in a row and then 3 posts in a row... I don't think i have ever seen that before.

and it would appear Daneel has never seen the edit button before :D

Daneel Olivaw
09-25-06, 08:06 PM
and it would appear Daneel has never seen the edit button before :DEdit is a bad bad button, it doesn't count on the posts total :p

crainger
09-25-06, 08:11 PM
I should have known...

Daneel Olivaw
09-25-06, 08:15 PM
3dmark2006 results

@stock (450/1320), 1.2V
3693 3dmarks

Oc'ed (520/1440), 1.2V
4071 3dmarks

Oc'ed (620/1440), 1.4V
4468 3dmarks

crainger
09-25-06, 08:22 PM
Volt modding. Its a dangerous game... :)

Daneel Olivaw
09-25-06, 08:41 PM
Volt modding. Its a dangerous game... :)True, but its a lot of fun and I checked with a voltmeter.