[TECH Report] As the second turns: the web digests our game testing methods

[Ibra]

Fixed. :thumbsup:

Keep dreaming.

http://gamegpu.ru/rpg/rollevye/of-orcs-and-men-test-gpu.html
http://gamegpu.ru/racing-simulators-/-gonki/project-cars-test-gpu.html
http://gamegpu.ru/mmorpg-/-onlayn-igry/shootmania-storm-test-gpu.html
http://gamegpu.ru/mmorpg-/-onlayn-igry/world-of-tanks-0820-test-gpu.html

These are non TWIMTBP/GE titles.

 

[ICDP]

Don't forget to turn up the voltage too!

At least AMD users have the option to play with voltage to allow them to attain much higher performance than would otherwise be possible. The GTX 660Ti is only matching a stock HD 7950 until the 7950 gets an overclock. Once that overclock is applied the 7950 even exceeds the GTX 680. I think my GTX 680 is a brilliant GPU but do miss the ability to overvolt to gain extra performance.

I have a HD 7950 that with an overclock of 1175/1550 matches or exceeds a GTX 680 overclocked and it does it with similar power consumption. An overclock on a HD 7950 will take no more than a few extra watts unless you seriously up the voltage. But the fact that any GPU cards allows the enthusiast to apply a voltage boost for higher overclocks should never be seen as a bad thing.

GTX 680 1.175 (1202 and VRAM +554)
HD 7950 1.170 (1175/1550)

http://hardforum.com/showpost.php?p=1039505601&postcount=66

Everyone is acting like 350-400 watts full system load is a massive amount, even 500 with a very large voltage boost is not a massive amount. Stop with this BS that because Tahiti uses slightly more power than Kepler it is a bad thing. It is still far below what previous the gen GTX 580 used.

GTX 660Ti vs HD 7950 system load according to HardOCP.

http://www.hardocp.com/article/2012/08/21/galaxy_geforce_gtx_660_ti_gc_3gb_overclocking_review/5

 

[Dark Shroud]

The GTX 660Ti is only matching a stock HD 7950 until the 7950 gets an overclock. Once that overclock is applied the 7950 even exceeds the GTX 680. I think my GTX 680 is a brilliant GPU but do miss the ability to overvolt to gain extra performance.

The over clocked 660 TI can catch up to a stock 7950.

This is something that's been frustrating and done since the 660 TI was released. NVidia shipped factory OC models to everyone who then just used older test data/drivers from AMD.

 

[Rikard]

Don't forget to turn up the voltage too!

No need. 1175/1450 MHz @ 0.95 V

 

[Whitestar127]

Here enjoy the benchies.
http://www.anandtech.com/bench/Product/647?vs=550

Turn up the clocks on Tahiti for more fun and it waves bye bye...

Hehe, allright. But the question was what those graphs in his post showed.

Maybe I should just have answered: "Some nice colors and a lot of numbers."

 

[notty22]

At least AMD users have the option to play with voltage to allow them to attain much higher performance than would otherwise be possible. The GTX 660Ti is only matching a stock HD 7950 until the 7950 gets an overclock. Once that overclock is applied the 7950 even exceeds the GTX 680. I think my GTX 680 is a brilliant GPU but do miss the ability to overvolt to gain extra performance.

Everyone is acting like 350-400 watts full system load is a massive amount, even 500 with a very large voltage boost is not a massive amount. Stop with this BS that because Tahiti uses slightly more power than Kepler it is a bad thing. It is still far below what previous the gen GTX 580 used.

GTX 660Ti vs HD 7950 system load according to HardOCP.

http://www.hardocp.com/article/2012/08/21/galaxy_geforce_gtx_660_ti_gc_3gb_overclocking_review/5

You state one thing, that a o/c 7950 can do this and not draw that much more power, then show Power usage( direct us to a image) of a o/c gtx 660ti (3GB model) vs a stock 7950.

 

[RaistlinZ]

Does anyone know if the latency issue only affected the 7950? Will 13.2's fix it for the entire 7xxx line, or just that one card?

 

[lavaheadache]

any GCN based card

 

[ICDP]

You state one thing, that a o/c 7950 can do this and not draw that much more power, then show Power usage( direct us to a image) of a o/c gtx 660ti (3GB model) vs a stock 7950.

I linked to that to show that the OC edition GTX 660Ti and HD 7950 (as used in the TR article referenced by the OP) use similar power. The only time a HD 7950 will consume significantly more power is when the voltage is increased to 1.2 volts or more. Even then it is still significantly less than previous gen GTX 580, so some perspective is required when judging power consumption.

Here is a list of power consumptions for various GTX 660Ti cards compared to a reference HD 7950. Under full load a reference GTX 660Ti consumes 29 watts less power than a HD 7950. Hardly a massive difference is it? This of course only applies to those who purchased stock clock 660Ti cards. If you purchased an OC edition then congrats, your card is drawing the same power as a HD 7950 but you have significantly reduced OC headroom.

http://www.hardwarecanucks.com/foru...roundup-asus-evga-gigabyte-galaxy-msi-22.html

At least there is that option on the HD 7950 to use more volts to attain significant overclocks. 800Mhz stock to an even moderate 7950 overclock of 1100Hz = 37.5% overclock. 40%-50% overclock on a HD 7950 leaves a GTX 660Ti in the dust, in fact it even exceeds a GTX 680 overclock as shown in my link. An overclocked GTX 660Ti does not even reach stock GTX 680 speeds for current gen games, let alone compete with an overclocked one.

 

[Imouto]

 

[MrK6]

I linked to that to show that the OC edition GTX 660Ti and HD 7950 (as used in the TR article referenced by the OP) use similar power. The only time a HD 7950 will consume significantly more power is when the voltage is increased to 1.2 volts or more. Even then it is still significantly less than previous gen GTX 580, so some perspective is required when judging power consumption.

Here is a list of power consumptions for various GTX 660Ti cards compared to a reference HD 7950. Under full load a reference GTX 660Ti consumes 29 watts less power than a HD 7950. Hardly a massive difference is it? This of course only applies to those who purchased stock clock 660Ti cards. If you purchased an OC edition then congrats, your card is drawing the same power as a HD 7950 but you have significantly reduced OC headroom.

http://www.hardwarecanucks.com/foru...roundup-asus-evga-gigabyte-galaxy-msi-22.html

At least there is that option on the HD 7950 to use more volts to attain significant overclocks. 800Mhz stock to an even moderate 7950 overclock of 1100Hz = 37.5% overclock. 40%-50% overclock on a HD 7950 leaves a GTX 660Ti in the dust, in fact it even exceeds a GTX 680 overclock as shown in my link. An overclocked GTX 660Ti does not even reach stock GTX 680 speeds for current gen games, let alone compete with an overclocked one.

This is more. The Kepler series simply isn't as power efficient as some parties would like everyone to believe. For example: http://www.techpowerup.com/reviews/VTX3D/Radeon_HD_7870_XT_Black/29.html . AMD and nvidia competitors are neck-and-neck for performance/watt at almost every tier, with nvidia starting to fall behind the higher the resolution gets. For the specific example being argued, coupled with increased overclocking ability and voltage options of the 7950 series card, AMD simply offers more flexibility when running their hardware.

 

[raghu78]

Will. Do you, or more to the point can you, notice anything in the TR graphs?I just want to see if you realize.

TR is not the only site on the web which does reviews.

http://www.hardocp.com/article/2013/01/01/powercolor_radeon_hd_7870_myst_edition_review/11

"Compared to the reference Radeon HD 7870 GHz edition, the PowerColor HD 7870 MYST represents a step up in performance from both the Radeon HD 7870 GHz edition and the GTX 660 Ti."

the GTX 660 Ti has a tough time beating a HD 7870 tahiti LE. forget a HD 7950 boost which is one step up and on par with GTX 670. With overclocking a HD 7950 (1150 mhz) beats a GTX 670(1250 Mhz) in the majority of games.

even now its not a contest between a GTX 660 Ti and HD 7950 boost. forget how its going to be with the new memory manager. you are fooling only yourself by saying a GTX 660 Ti competes with a HD 7950 boost :thumbsup:

 

[BrightCandle]

Comparison of a 680 and a 7970 Ghz Edition looking at smoothness.
http://alienbabeltech.com/main/?p=33257

From the conclusion

From just looking at eleven benchmarks including nine games, we see that the HD 7970 at GHz edition clocks has more difficulty providing even frame rates. The GTX 680 does a better job of delivering a smoother experience in our benchmarks despite the HD 7970 generally having higher fps (framerates). So far, it appears that Scott Wasson of the Tech Report was correct that the Radeon’s GCN Catalyst drivers have difficulty with smoothness, especially with some newer games.

 

[raghu78]

Comparison of a 680 and a 7970 Ghz Edition looking at smoothness.
http://alienbabeltech.com/main/?p=33257

From the conclusion

let them test battlefield 3 with 310.70 and 310.90 whql drivers and show the Nvidia GTX 680 providing a smoother experience. that driver has proved troublesome with significant stuttering and frame latency problems. here are users complaining of stuttering issues in bf3 on gtx 600 cards. one user has put up detailed frametime charts

http://www.overclock.net/t/1339698/...is-better-hd-7870-vs-660-ti-benchmarks-inside

http://www.overclock.net/t/1346657/battlefield-3-unplayable-on-r310-drivers

and given that AMD has now clearly shown 13.2 beta ( which is due this week but a leaked version is already being used by many users) fixing frame latency problems in dx9 games like skyrim, borderlands 2 and guild wars 2 and also stated that the new memory manager due next month will resolve frame latency issues for DX10 / DX11 games its no more a major issue.

nvidia on the contrary has regressed in bf3 and needs to correct the situation. drivers are not perfect with either side. with multi gpu though nvidia does have the edge with their frame metering and more robust SLI solution.

 

[Will Robinson]

Quoting ABT in any serious forum is like using Charlie at S/A as a source for NVDA problems.

The wounds run deep it seems with Appopin and Charles.

 

[Rikard]

Quoting ABT in any serious forum is like using Charlie at S/A as a source for NVDA problems.

The wounds run deep it seems with Appopin and Charles.

I thought ABT is were you go when you have been banned from every other forum on the Internet. Sort of like VC&G heaven. Or hell.

 

[Lonbjerg]

Quoting ABT in any serious forum is like using Charlie at S/A as a source for NVDA problems.

The wounds run deep it seems with Appopin and Charles.

Was that a irrelevant stab at ABT...or a denial of microstutter?

 

[Homeles]

Was that a irrelevant stab at ABT...or a denial of microstutter?

Why not both?

 

[3DVagabond]

ABT, unfortunately, has little credibility left after apoppin's rants about AMD, this forum, and it's members.

 

[Lonbjerg]

ABT, unfortunately, has little credibility left after apoppin's rants about AMD, this forum, and it's members.

I can ask you the same thing:

Was that a irrelevant stab at ABT...or a denial of microstutter?

 

[3DVagabond]

I can ask you the same thing:

Was that a irrelevant stab at ABT...or a denial of microstutter?

It is what it is. No secret message. Just read the words and let them sink in. They're simple words. Nothing that needs any interpreting.

 

[Lonbjerg]

It is what it is. No secret message. Just read the words and let them sink in. They're simple words. Nothing that needs any interpreting.

You know when people have an "agenda"...when they cannot answer with a simple yes or no.

Let me make it more easy for you:

Do you DENY microstutter?

Simple yes/no.

 

[jackstar7]

You know when people have an "agenda"...when they cannot answer with a simple yes or no.

Let me make it more easy for you:

Do you DENY microstutter?

Simple yes/no.

Does anyone at this point? Both "houses" are plagued by it to various degrees.

Also, why harp on this point?

 

[KingFatty]

"... I want the truth!"

"you can't handle the truth!"

 

[3DVagabond]

You know when people have an "agenda"...when they cannot answer with a simple yes or no.

Let me make it more easy for you:

Do you DENY microstutter?

Simple yes/no.

What does my comment about ABT have to do with the existence of microstutter? Let me clarify my statement. apoppin has an agenda. Therefor, nothing he says can be relied upon.

As far as your deflection about "microstutter".

1, I'd like to see blind testing to see at what point the measured frame latencies are visible in real world use.
Reason? We've seen larger amounts of latency measured with a 6970, 560ti, and 470 with never a report from testers or owners that they have visible stuttering issues. These results weren't just in a couple of titles either. They were pretty consistent across the titles tested.

2, I'd like to see this testing done in a larger selection of games to see how wide spread the latency issue is.
Reason? They used different games in different reviews and had different results.

Are there measurable differences in latency between the cards tested in the games tested? Yes. I'm not going to take that limited information and apply a blanket statement from it. No.