Bench scores for 7800GTX 512MB

I have been using ATI for as long as i can remember and i had a 1800xt on order for well over 2 weeks. Never showed up and i got confirmed that i can pick up the 7800 gtx 512 tomorrow if i want (5 of them in my local store). That made my choice easier. I'm gonna be dissapointed if the GTX 512 doesn't do 10K/20K with a MILD overclock. Even the mobile 7800 does pretty good. I benched my lappy today with the new drivers (drivers had to be modified to work in this Dell)

Total stock, Dothan at 2ghz.

http://www.danghardware.com/3dlappysmall.jpg

A bit off topic, but it shows what we can expect.

i think the x1800 cards do well in dx 9 situations. 03 is not fully dx 9, so I think thats why the GTX leads.

another thing, considering the the XT gets only a few hundred points less than the 512MB GTX even though its pixel fillrate is about 30% less (10K vs 14K), says alot about how efficient those 16 pipes really are. If nothing else, the x1800 cards should be applauded for their extreme efficiency.

Quote:

---

Originally Posted by gunit

i think the x1800 cards do well in dx 9 situations. 03 is not fully dx 9, so I think thats why the GTX leads.

another thing, considering the the XT gets only a few hundred points less than the 512MB GTX even though its pixel fillrate is about 30% less (10K vs 14K), says alot about how efficient those 16 pipes really are. If nothing else, the x1800 cards should be applauded for their extreme efficiency.

---

I say nothing......I don't care about 3DMark..it's not a game...and it's 1024x768....woohoo... :p:

I'll wait for the reviews thank you very much :D

Quote:

---

Originally Posted by DilTech

Hicks, get your facts straight.
XT = wins in 01 and 05
GTX = Wins in 03
GTX 512 = Wins in 03 and 05
Any questions?

---

Get your facts straight. Never did i say the XT was better, just was saying it doesn't get 16k most of the time it's 17k with a decent CPU.

And a normal GTX never beats a XT in 03, stock of course.

here are some benches from nvnews

Its looking good.

What venders will be selling them on Monday? BFG, PNY, EVGA ect.. Or just XFX? I don't really care for XFX, but..

Quote:

---

Originally Posted by Hicks

Get your facts straight. Never did i say the XT was better, just was saying it doesn't get 16k most of the time it's 17k with a decent CPU.

And a normal GTX never beats a XT in 03, stock of course.

---

Now that I look again, you're right, the XT beats the 7800gtx by less than 1,000 points... Looks like the original review I looked at used a dual-core cpu, which was enough for the 7800gtx to take that victory.

Well Monarch has the BFG and eVGA versions up now as well. All at the same price. Im waiting til monday to order, probably from the egg. Im curious to see how well this goes with my 3800+ X2 and my new DFi nF4 SLI Expert board.

Quote:

---

Originally Posted by zakelwe

Ati's pipes do more work per pipe but even so.

---

Not true. A 7800GTX at equal pipeline count and clockspeed is faster than a X1800XT as proven in a driverheaven article.

Quote:

---

Originally Posted by gunit

another thing, considering the the XT gets only a few hundred points less than the 512MB GTX even though its pixel fillrate is about 30% less (10K vs 14K), says alot about how efficient those 16 pipes really are. If nothing else, the x1800 cards should be applauded for their extreme efficiency.

---

Also not true. Fillrate is Core clockspeed x # or ROPs. The 7800GTX only has 16 ROPs so it's fillrate is identicle to a X1800XT at the same clockspeed.

Quote:

---

Originally Posted by Flak Monkey

Also not true. Fillrate is Core clockspeed x # or ROPs. The 7800GTX only has 16 ROPs so it's fillrate is identicle to a X1800XT at the same clockspeed.

---

not true, to use your favorite words. its not that simple. the GTXs fill rate is like 16x1.5=24. al you have to do is check fill rate test in 01, 03, or 05. A stock 512mb GTX will have near 14K mp/sec (580x24). A stock XT will have near 10K mp/sec, (625x16).

and, a GTX @ 625 like and XT will NOT have identical fill rate. wrong wrong wrong. It will have near 15K mp/sec, (625x24). Dont take my word for it, just check any 3dmark test with fillrate.

oh here ill make it easy for. according to your 16 ROP theory, at 527 core the GTX should have around 8,000 mp/sec. It doesnt,

http://service.futuremark.com/compare?2k3=4392359

its close to 12,600 mp/sec, as it should be. 24x527.

If I test my x1800 at 527 core, fill rate will be around 8,000.

Quote:

---

Originally Posted by Flak Monkey

Not true. A 7800GTX at equal pipeline count and clockspeed is faster than a X1800XT as proven in a driverheaven article.

Also not true. Fillrate is Core clockspeed x # or ROPs. The 7800GTX only has 16 ROPs so it's fillrate is identicle to a X1800XT at the same clockspeed.

---

In theory yes...

And well since its all lies then why is it that a X1800XT 16 pipe is only very much slower than a 24 pipe 7800GTX... And im talking 256vs256...

Edit: Just checked the orb and in the 3dmark fill rate tests 7800GTX seems to be the 30% faster... but in the tests its a little slower or a little faster...

ROPs x Clock speed is fill rate, that is a fact not a theory. Email nvidia if you don't believe me. Plus your looking at the wrong test. You should be looking at single-texturing fill rate. If you look at that you will see that your numbers and my number are both way off. This is because these cards can't match their theoretical fill rate in the real world. Multi-texturing is very close to the pipeling x clock speed your completely saturating the ROPs and are able to get more data out but this is not actual fill rate.

BTW there was no need to mock me with that first sentence. My intention was not to be rude. Just to correct some misinformation.

Quote:

---

Originally Posted by Flak Monkey

ROPs x Clock speed is fill rate, that is a fact not a theory. Email nvidia if you don't believe me. Plus your looking at the wrong test. You should be looking at single-texturing fill rate. If you look at that you will see that your numbers and my number are both way off. This is because these cards can't match their theoretical fill rate in the real world. Multi-texturing is very close to the pipeling x clock speed your completely saturating the ROPs and are able to get more data out but this is not actual fill rate.

BTW there was no need to mock me with that first sentence. My intention was not to be rude. Just to correct some misinformation.

---

point is what you said was wrong and I had to correct it. identicaly clocked GTX and x1800 will NOT have same fill rate.

They will have the same fill rate. Just because one is able to do more when using more than one texture doens't change the fact that it's true fill rate is ROPs x Clocks.

Quote:

---

Originally Posted by gunit

not true, to use your favorite words. its not that simple. the GTXs fill rate is like 16x1.5=24. al you have to do is check fill rate test in 01, 03, or 05. A stock 512mb GTX will have near 14K mp/sec (580x24). A stock XT will have near 10K mp/sec, (625x16).

and, a GTX @ 625 like and XT will NOT have identical fill rate. wrong wrong wrong. It will have near 15K mp/sec, (625x24). Dont take my word for it, just check any 3dmark test with fillrate.

oh here ill make it easy for. according to your 16 ROP theory, at 527 core the GTX should have around 8,000 mp/sec. It doesnt,

http://service.futuremark.com/compare?2k3=4392359

its close to 12,600 mp/sec, as it should be. 24x527.

If I test my x1800 at 527 core, fill rate will be around 8,000.

---

That link shows texture fillrate. That function's contained within the main shader pipelines, 24 of which are available.

There are 16 ROPs. The ROPs control the pixel fillrate.

A stock GTX will have a texture fillrate of 10320 mega-texels per second, and a pixel fillrate of 6880 megapixels per second.

Quote:

---

Originally Posted by MaxxxRacer

As much as I dislike nVidia and as much as I like ATI, I must say nVidia has been owning the high end market for a long time now and will continue to do so for a while longer.

---

I guess we'll be holding out hope that the r580 will come to market the same time as nvidia's next product and beats nvidia... well I'm hoping.

Quote:

---

Originally Posted by Unoid

I guess we'll be holding out hope that the r580 will come to market the same time as nvidia's next product and beats nvidia... well I'm hoping.

---

:p:
possible but unlikely

Quote:

---

Originally Posted by Flak Monkey

Not true. A 7800GTX at equal pipeline count and clockspeed is faster than a X1800XT as proven in a driverheaven article.

---

But it's not the same clock speed, to get the score it does the Ati 16 pipes work faster and so the Ati does more work per pipe than the nvidia to get the similar score. I am not talking efficiency here I am talking work done per time period ..please re-read and re think what I have written.

Think of it like this, a 4 cylinder engine at 7000rpm produces 200bhp, it thus produces more work per cylinder than a 6 cylinder engine which produces 200bhp at 4000rpm.

Hmm, on second thoughts I think my wording was poor. Let me try again.

Ati have chosen reduced pipes and higher clocks so resulting in those pipes having to work harder to produce a similar or higher score than the nvidia choice of more and slower.

That's what I am trying to say


Regards


Andy

Who cares ? Whatever takes the orb down is fastest (here).
Now where is Sampsa ?

Quote:

---

Originally Posted by Ubermann

Who cares ? Whatever takes the orb down is fastest (here).
Now where is Sampsa ?

---

Not really.....whatever card beats every other card in real life gaming is the fastest....that's my vision on it anyway...

3DMark is so overrated....

Thats wrong forum to say 3DMark is overrated :D
GTX 512 SLI scores 12,549 in 05 on Fx-53 @ 2580
Trying to find some more info and i hope he spend money on a better CPU cuz that score SUX! =)
http://www.hardforum.com/showpost.ph...4&postcount=57

I've tested now GeForce 7800 GTX 512 MB, GeForce 7800 GTX 256 MB and Radeon X1800 XT 512 MB with Doom 3, Half-Life 2, Battlefield 2, Call of Duty 2, Quake 4 and F.E.A.R.

New GeForce 7800 "Ultra" (It says GF-7800-U on GPU) beats Radeon X1800 XT 512 MB in every test I've ran so far (Default reference clocks).

Quote:

---

Originally Posted by Ubermann

Thats wrong forum to say 3DMark is overrated :D
GTX 512 SLI scores 12,549 in 05 on Fx-53 @ 2580
Trying to find some more info and i hope he spend money on a better CPU cuz that score SUX! =)
http://www.hardforum.com/showpost.ph...4&postcount=57

---

When changing from single-card to SLI the CPU becomes bottleneck.

Quote:

---

Originally Posted by Sampsa

I've tested now GeForce 7800 GTX 512 MB, GeForce 7800 GTX 256 MB and Radeon X1800 XT 512 MB with Doom 3, Half-Life 2, Battlefield 2, Call of Duty 2, Quake 4 and F.E.A.R.

New GeForce 7800 "Ultra" (It says GF-7800-U on GPU) beats Radeon X1800 XT 512 MB in every test I've ran so far (Default reference clocks).

---

Sampsa, It is being said that ATI is increasing clocks of the XT series to 650 775. Could you tell us how that will change the results in the games listed above?