6800GT oc'ing.... better

[SickBeast]

Originally posted by: GeneralGrievous

Where have you seen an x800 Pro that could beat out an XT? Even if it's clocked higher, it has 4 less pipelines. I would be very surprised to see one beat out an XT.

I didn't mean that it could beat an XT in raw power, rather that they are acheiving XT or higher clock speeds.

Well, a 6800GT CAN beat out an Ultra when overclocked, so that makes it the better card IMO. You seem to think that it is only marginally better than an X800PRO, which I disagree with. I think it's clearly the better purchase.

 

[Regs]

CPU limited with battle field? Could try another game.

And where did you find a 6800GT for sale? lol

Edit, NM. Sucker is over 500 dollars! Ive been behind.

 

[UnTech]

Originally posted by: Regs
CPU limited with battle field? Could try another game.

And where did you find a 6800GT for sale? lol

Edit, NM. Sucker is over 500 dollars! Ive been behind.

I think Evga's website has them for MSRP $399 + Far Cry.

 

[DKlein]

Fry's $399

 

[Regs]

Ah ok. I freaked out when I saw neweggs price of a 6800 GT.

I think some games are more CPU bound then GPU bound which will change in the future. Even if you have a 2.6 GHz A64 or 3.6EE, you are not going to brake a barrier with some of these games. Plus, some games were just not designed to be FPS demons.

 

[Concillian]

Originally posted by: Regs
CPU limited with battle field? Could try another game.

BFV is probably not CPU bound. If you check the link to my benchmarks I posted much earlier in the thread you'll see that I got the same frame rates on my AXP at 1400 MHz as I did at 2200 MHz. That was with a 5900XT, and a 6800GT has more of a chance of becoming CPU bound, but I was getting minimum frame rates of over 30 FPS at 1400 MHz, I highly doubt he's getting too limited by the CPU at that res and visual quality with his processor setup.

 

[Acanthus]

Id probably run 16x12 2x8x

At 16x12 with 2x AA there should be no aliasing at all.

 

[ponyo]

If mine can hit Ultra speed when I receive my BFG 6800GT I will be very happy.

 

[GeneralGrievous]

Well, a 6800GT CAN beat out an Ultra when overclocked, so that makes it the better card IMO. You seem to think that it is only marginally better than an X800PRO, which I disagree with. I think it's clearly the better purchase.

It is. But looking at the firingsquad benches the only title the GT really has a substantial lead in is call of duty. Set aside that game and the GT wins the rest of the benchmarks by an average of around 5%. Better? Certainly, at stock speeds. But going from 350 core to 400 is only a 14% overclock, and a 10% overclock on the memory. If I can obtain a 20%+ overclock on the x800 Pro (570 core, which isn't unreasonable, if I can get an XT to 585), it's likely to be the faster card of the 2, that 5% advantage is now in ATI's favor. That said, I would still buy a GT over the Pro (giving up 5% performance is an acceptable cost for the extra features), but if you can obtain a Pro for cheaper that decision is not as clear. The x800 pros have 2.0 ns chips rated at 500 mhz that also seems to be overclocking extremely well.

 

[SickBeast]

Originally posted by: GeneralGrievous

Well, a 6800GT CAN beat out an Ultra when overclocked, so that makes it the better card IMO. You seem to think that it is only marginally better than an X800PRO, which I disagree with. I think it's clearly the better purchase.

It is. But looking at the firingsquad benches the only title the GT really has a substantial lead in is call of duty. Set aside that game and the GT wins the rest of the benchmarks by an average of around 5%. Better? Certainly, at stock speeds. But going from 350 core to 400 is only a 14% overclock, and a 10% overclock on the memory. If I can obtain a 20%+ overclock on the x800 Pro (570 core, which isn't unreasonable, if I can get an XT to 585), it's likely to be the faster card of the 2, that 5% advantage is now in ATI's favor. That said, I would still buy a GT over the Pro (giving up 5% performance is an acceptable cost for the extra features), but if you can obtain a Pro for cheaper that decision is not as clear. The x800 pros have 2.0 ns chips rated at 500 mhz that also seems to be overclocking extremely well.

I have yet to see an overclocked X800PRO that can beat out an overclocked 6800GT. Where you're getting this 5% figure (in ATi's favor) from is beyond me; I don't think it exists.

 

[GeneralGrievous]

We have yet to see any real benches of an overclocked pro. That 5% figure is mostly theoratical based on math.

 

[SickBeast]

Originally posted by: GeneralGrievous
We have yet to see any real benches of an overclocked pro. That 5% figure is mostly theoratical based on math.

Ok, so you're saying that the X800PRO can overclock to 570 on the core. That's great, but to equal the X800XT it would have to be clocked 25% higher due to the lack of pipelines, so long as the memory speeds are the same (which they should be capable of doing since they both use the same RAM chips). 570/500 is only 14% faster than the XT, and that doesn't take into account the fact that the XT can also be overclocked.

Face it, you're comparing a crippled card to one with 16 pipelines that it doesn't have much chance competing against. 6800GT all the way. I'm surprised you're even debating this.

 

[GeneralGrievous]

Ok, so you're saying that the X800PRO can overclock to 570 on the core. That's great, but to equal the X800XT it would have to be clocked 25% higher due to the lack of pipelines, so long as the memory speeds are the same (which they should be capable of doing since they both use the same RAM chips). 570/500 is only 14% faster than the XT, and that doesn't take into account the fact that the XT can also be overclocked.

Why does the Pro have to be equal to an XT in performance? I thought we were comparing the Pro to the GT. The XT is already above GT and Pro performance and is at a higher MSRP; it has nothing to do with the discussion.

Face it, you're comparing a crippled card to one with 16 pipelines that it doesn't have much chance competing against. 6800GT all the way.

Look at the benches I linked to earlier. I don't know how an x800 Pro can be crippled when it runs neck and neck with the GT in half of them.

 

[Pete]

There will be an almost ideal opportunity to compare the NV40's and R420's core temps directly, as each will have a card sold with the new Arctic Cooling VGA Silencer v2 (HIS and Sapphire for ATi, and maybe PNY or Prolink or G-something for nVidia). So hopefully a website will snag both and settle this seemingly minor issue (if both cards use the same power, it stands to reason they'll release about the same heat, considering they're both 130nm and use GDDR3 [ignoring low-K]).

But, as is, I'm not sure you can compare temps b/w a single-slot X800XT and a double-wide 6800U. Comparing a X800P/X800XT and a 6800GT seems fairer, though.

 

[DKlein]

Now that my room temp has gone down a good bit, I've started overclocking agian. Coolbits let me get to 1.19, but no further. I've managed to get 3DMark to run a number of times now at 411/1.19 (~12000), but I can't loop it so I'm in need of a good torture tester.

 

[Jeff7181]

Originally posted by: Dman877
IS BFV a demanding game with high quality graphics (like Farcry) or is it just poorly coded? The screenshots I've seen don't look that special and I've never played any of the BF series (I do have BF1942 though, just never installed /shrug).

It's basically BF1942 with better textures. No real improvements in the game itself... I don't own the game for obvious reasons (if you read my sig) but I just got back from a friend's place and it does look better than BF1942, but that's not saying a whole lot as I wasn't impressed with the quality of the graphics in BF1942.

 

[SickBeast]

Originally posted by: GeneralGrievous

Ok, so you're saying that the X800PRO can overclock to 570 on the core. That's great, but to equal the X800XT it would have to be clocked 25% higher due to the lack of pipelines, so long as the memory speeds are the same (which they should be capable of doing since they both use the same RAM chips). 570/500 is only 14% faster than the XT, and that doesn't take into account the fact that the XT can also be overclocked.

Why does the Pro have to be equal to an XT in performance? I thought we were comparing the Pro to the GT. The XT is already above GT and Pro performance and is at a higher MSRP; it has nothing to do with the discussion.

Face it, you're comparing a crippled card to one with 16 pipelines that it doesn't have much chance competing against. 6800GT all the way.

Look at the benches I linked to earlier. I don't know how an x800 Pro can be crippled when it runs neck and neck with the GT in half of them.

Not when the GT is overclocked, or even when both the GT and the X800PRO are overclocked. I don't believe this, you don't think the GT is superior. You're acting pigheaded.

 

[Acanthus]

Originally posted by: Jeff7181

Originally posted by: Dman877
IS BFV a demanding game with high quality graphics (like Farcry) or is it just poorly coded? The screenshots I've seen don't look that special and I've never played any of the BF series (I do have BF1942 though, just never installed /shrug).

It's basically BF1942 with better textures. No real improvements in the game itself... I don't own the game for obvious reasons (if you read my sig) but I just got back from a friend's place and it does look better than BF1942, but that's not saying a whole lot as I wasn't impressed with the quality of the graphics in BF1942.

Bot AI is significantly improved (further improved with 1.10), it was the main flaw with BF1942 imho.

 

[ss284]

Originally posted by: SickBeast

Originally posted by: GeneralGrievous

Ok, so you're saying that the X800PRO can overclock to 570 on the core. That's great, but to equal the X800XT it would have to be clocked 25% higher due to the lack of pipelines, so long as the memory speeds are the same (which they should be capable of doing since they both use the same RAM chips). 570/500 is only 14% faster than the XT, and that doesn't take into account the fact that the XT can also be overclocked.

Why does the Pro have to be equal to an XT in performance? I thought we were comparing the Pro to the GT. The XT is already above GT and Pro performance and is at a higher MSRP; it has nothing to do with the discussion.

Face it, you're comparing a crippled card to one with 16 pipelines that it doesn't have much chance competing against. 6800GT all the way.

Look at the benches I linked to earlier. I don't know how an x800 Pro can be crippled when it runs neck and neck with the GT in half of them.

Not when the GT is overclocked, or even when both the GT and the X800PRO are overclocked. I don't believe this, you don't think the GT is superior. You're acting pigheaded.

Was this statement supposed to have any sort of meaning or value to it whatsoever or are you just spouting nonsense because you have nothing better to say? GeneralGrevious's statements all have some logical merit to them, while yours dont. You dont even bother to read what other people are saying. Yes, the 6800 gt is faster in benchmarks. All he is saying is that since percentage overclocks on 800pros seem to be better, the performance of the two when overclocked are neck and neck.

-Steve

 

[Concillian]

Originally posted by: Acanthus
Bot AI is significantly improved (further improved with 1.10), it was the main flaw with BF1942 imho.

I like BFV, I play it a lot. But BFV is NOT a single player game.

Bot AI is improved over BF1942, but bot AI in BF1942 was, quite literally, the worst I've ever seen. The improvements in BFV are there, but they don't help a whole lot.

I feel sorry for anyone who boutght BFV for a single player game, because the SP is pretty bad. You don't even get the campaign like in BF1942, you just pick maps and go.

 

[SickBeast]

Originally posted by: ss284

Originally posted by: SickBeast

Originally posted by: GeneralGrievous

Ok, so you're saying that the X800PRO can overclock to 570 on the core. That's great, but to equal the X800XT it would have to be clocked 25% higher due to the lack of pipelines, so long as the memory speeds are the same (which they should be capable of doing since they both use the same RAM chips). 570/500 is only 14% faster than the XT, and that doesn't take into account the fact that the XT can also be overclocked.

Why does the Pro have to be equal to an XT in performance? I thought we were comparing the Pro to the GT. The XT is already above GT and Pro performance and is at a higher MSRP; it has nothing to do with the discussion.

Face it, you're comparing a crippled card to one with 16 pipelines that it doesn't have much chance competing against. 6800GT all the way.

Look at the benches I linked to earlier. I don't know how an x800 Pro can be crippled when it runs neck and neck with the GT in half of them.

Not when the GT is overclocked, or even when both the GT and the X800PRO are overclocked. I don't believe this, you don't think the GT is superior. You're acting pigheaded.

Was this statement supposed to have any sort of meaning or value to it whatsoever or are you just spouting nonsense because you have nothing better to say? GeneralGrevious's statements all have some logical merit to them, while yours dont. You dont even bother to read what other people are saying. Yes, the 6800 gt is faster in benchmarks. All he is saying is that since percentage overclocks on 800pros seem to be better, the performance of the two when overclocked are neck and neck.

-Steve

Well, aside from the pig head comment, it had value and was not nonesense. Look, the 6800GT is basically an Ultra or better when overclocked, that much is obvious. As I pointed out, the X800PRO would have to be clocked at least 25% higher on the core than an XT in order for it to compete. That simply is not happening. That would require a core speed of 625mhz. The best I've seen on one of those cards is 580mhz. He even went to the point of saying that the X800PRO isn't crippled; what does that tell you? To me it shows blatent personal bias. Think what you will, but my comments were much more logical and well thought out than his were.

 

[GeneralGrievous]

The XT is faster than the Ultra, at least according to most review sites around.

To me it shows blatent personal bias. Think what you will, but my comments were much more logical and well thought out than his were.

Right. You must REALLY place high importance in a 5% performance advantage. :roll:

 

[SickBeast]

Originally posted by: GeneralGrievous
The XT is faster than the Ultra, at least according to most review sites around.

To me it shows blatent personal bias. Think what you will, but my comments were much more logical and well thought out than his were.

Right. You must REALLY place high importance in a 5% performance advantage. :roll:

It's not 5% at all. You keep coming up with that number off the top of your head. It's more like 15-20%, which is substantial.

My numbers are perfectly logical; the best an X800PRO can do is a 14% overclock on the core, which is 11% shy of the 25% it needs to overtake the XT. This shows me that it's 11% slower than the XT, and that doesn't even take into consideration that the XT can also be overclocked. The GT has been shown to be able to easily clock to Ultra speeds.

Now, you say that the XT is faster than the Ultra, which is not the general consensus AFAIK. Most people seem to accept the fact that the two cards are equal. If you look at a more recent review using new drivers, you will see that the 6800U even soundly beats the XT in Far Cry, which was initially the main reason for choosing the XT over the Ultra.

Believe what you will, but I feel that my numbers show a definite 11% advantage for the GT, and that is a worst-case scenario. Chances are it can overclock beyond 400mhz on the core.

 

[Illissius]

I've lately been skipping to the overclocking section on every review with an index, as well as reading forums posts, and it seems to me that in general, GTs overclock to 410-420MHz, while X800 Pros to 520-540. Let's examine that in terms of fill rate:

350 * 16 = 5600
475 * 12 = 5700

At stock speeds, the X800 Pro has a slight advantage.

410 * 16 = 6560
520 * 12 = 6240
420 * 16 = 6720
540 * 12 = 6480
460 * 16 = 7360
570 * 12 = 6840
(last two added for those who decide based on the "what if I'm unreasonably lucky and achieve the highest overclocks I've heard of anywhere" factor)

As you can see, as you raise clock speeds, the situation reverses, and the 6800 GT starts pulling away from the X800 Pro. This is because it has 33% more pipes, and thus gains 33% more fillrate per extra MHz. If we take a look at other factors pertaining to performance:

6800GT:
SM 3.0
Much better OpenGL performance
Somewhat higher performance per clock
Drivers and games not yet (fully) optimized for it -> plenty of room for additional improvement
Will with 99% probability be faster at Doom 3 and other games using the engine

X800 Pro:
3Dc
Will with 99% probability be faster at HL2 and other games using the engine

Thus, I believe we can safely conclude the 6800GT to be the victor, other non-performance-related things (such as power draw, noise, heat, Linux support, brand preference, etc.), aside.

Anything I missed?

 

[SickBeast]

You can't base an evaluation on fillrate alone. The GT has lower fillrate, but gets much more done per clock cycle (as you noted). The more logical thing to do would be to post benchmarks with each card overclocked to the max, but I have yet to find such numbers on the web.

I agree with your conclusion, but we really should be basing these things from benchmarks rather than our own personal evaluations.