[SugarCoat]

Was looking around for a review when it launched. Tried again and found one. For anyone interested in what 630MHz Core 2000MHz does over the stock 575/1800MHz. I want to mention that i'm 99% positive the actual "Stream Processor" clocks are the exact same. 1.35GHz. This goes for many of the 8800GTX OC cards i've seen coming out, except for the BFG OC which specifically lists theirs as clocked at 1400MHz. Their memory speed is still stock on that card, so i'll enjoy a comparison of that against the XXX or the Evga Superclocked, or a vanilla card for that matter, once these cards get out there to see how the shader clock effect performance.

http://www.motherboards.org/reviews/...re/1685_1.html

[Bludd]

Are those graphs correct? The XXX version is slower than the non-XXX in almost everything? And it costs more? And is clocked higher? Err...

[silent_guy]

Originally Posted by SugarCoat

I want to mention that i'm 99% positive the actual "Stream Processor" clocks are the exact same. 1.35GHz.

I've been wondering about that, but it's never mentioned. By now, I would have expected there to be some kind of hack tools available to read out the real clock, but I haven't seen any. What makes you so sure?

[SugarCoat]

Originally Posted by Bludd

Are those graphs correct? The XXX version is slower than the non-XXX in almost everything? And it costs more? And is clocked higher? Err...

i have to admit i didnt even notice that. I just skimmed them assuming the top bar was the OC....great.

CPU limitation and a margine for error caused by fraps problably? They dont look mixed up since the XXX OC tops out when the resolution is maxed in pretty much all those benchmarks.

Originally Posted by silent_guy

I've been wondering about that, but it's never mentioned. By now, I would have expected there to be some kind of hack tools available to read out the real clock, but I haven't seen any. What makes you so sure?

Well two things. One you already noticed, its not listed! If it was they'd boast about it. The second is that if you do to XFX's site they have a spec called the "Core clock" posted for their vanilla GTX and the GTX XXX and they're the same. Its seperate of the "Clock Rate" spec which is also listed. To be honest the most blatent proof is they dont boast it if it was faster which they surely would. Raw numbers sell these cards at the high costs, they know that.

[Monty]

Rivatuner can read the clock domains. Ive clocked my 8800gtx to 702 core, the shaders are around 1.620ghz at that clock .

Havnt got around to testing it yet though

[weeds]

Originally Posted by silent_guy

I've been wondering about that, but it's never mentioned. By now, I would have expected there to be some kind of hack tools available to read out the real clock, but I haven't seen any. What makes you so sure?

The Stream Processor clocks for XFX 8800GTX XXX edition are indeed 1350,
The EVGA 8800GTX KO ACS³ are 1458.

XFX 8800GTX XXX


EVGA 8800GTX KO ACS³

[Bludd]

Originally Posted by SugarCoat

i have to admit i didnt even notice that. I just skimmed them assuming the top bar was the OC....great.

CPU limitation and a margine for error caused by fraps problably? They dont look mixed up since the XXX OC tops out when the resolution is maxed in pretty much all those benchmarks.

Yeah, but it loses in so many others. It is strange. Maybe someone with a deeper insight than me will be able to explain it.

[weeds]

Originally Posted by Bludd

Yeah, but it loses in so many others. It is strange. Maybe someone with a deeper insight than me will be able to explain it.

Maybe its just the review, GURU3D's review it isnt slower.
http://www.guru3d.com/article/Videocards/412/1/

[AlexV]

I think they mixed up the charts, no biggie

[ben6]

Hi,

Couple of things,

1. I ran the tests with a clean system.
2. With FRAPs there's always a bit of inexactedness, especially if you're manually doing a walkthrough of a whole level of a game like Tomb Raider or a whole track in Need For Speed Carbon. The starting point of the FRAPs run can vary and the ending point slightly as well. I noticed the variations, but attributed them to trying to duplicate as close as possible a walkthrough of an entire level, with things like loading times for cut scenes and higher framerate when I stopped. It might have been a fairer look to show the entire FRAPs graph for each benchmark, but I can't do that on that site
3. You'll notice that the scores with AA+AF show a far different story as to framerate with the OC card consistently outperforming the regular edition.
4. I reran the tests several times

[ben6]

I should state that each runthrough is a complete level. For example in Tomb Raider Legend I start the FRAPs run at the beginning of the first level to the temple with the revolving spike trap in the first part of the temple. With all of the jumps and variations in how I kill enemies I'm surprised there isn't more variation in framerate in the game. I try to be as exactly the same each time, but there is a variation of up to 5%-10% in framerate sometimes more.

I wish I could do it exactly the same but I generally am playing the games exactly as I would if I bought the game for the first time with FRAPs running in the background. Then I start the level over and finish the level three total times. If framerate is more or less consistent I use the highest framerate from those three runs.