5770 benchmarks at AlienBabelTech

[bryanW1995]

benchmarks: http://alienbabeltech.com/main/?p=12436



edit: here's the bottlenecking article: http://alienbabeltech.com/main/?p=12474

and here's the IQ analysis: http://alienbabeltech.com/main/?p=12648

great job bfg!

 

[soccerballtux]

hm, that's an impressive card.

 

[BFG10K]

Thank goodness someone linked to it on their own accord in a post, because I couldn?t do it (against the TOS of the forum).

I also tested bottlenecking too and while I can?t link to it in my posts either, I found the 5770 is actually primarily limited by its core, not by its memory, as is commonly accepted.

Actually the two were quite close because it?s quite a balanced part, but the core edged it overall. From a 20% underlock on each, the core lost 11.20% performance overall, while the memory lost 8.46%.

 

[toyota]

Originally posted by: BFG10K
Thank goodness someone linked to it on their own accord in a post, because I couldn?t do it (against the TOS of the forum).

I also tested bottlenecking too and while I can?t link to it in my posts either, I found the 5770 is actually primarily limited by its core, not by its memory, as is commonly accepted.

Actually the two were quite close because it?s quite a balanced part, but the core edged it overall. From a 20% underlock on each, the core lost 11.20% performance overall, while the memory lost 8.46%.

if thats true then how do you explain it losing to the 4870 and 4890? no matter how high you overclock the 5770 it cant touch a stock clocked 4890 in some games. http://img360.imageshack.us/im.../1406/asushd577016.jpg

 

[toyota]

Originally posted by: soccerballtux
hm, that's an impressive card.

you do realize the card came out 10 days ago and benchmarks are everywhere?

 

[BFG10K]

Originally posted by: toyota

if thats true then how do you explain it losing to the 4870 and 4890? no matter how high you overclock the 5770 it cant touch a stock clocked 4890 in some games.

We can rule out the memory because it doesn?t seem to be the primary limitation based on my findings. There?s a bottleneck else where, and I?m convinced it?s at the driver level, intentional or not.

I don?t think we?ve seen the full performance of the 5xxx series yet, and I expect something interesting to happen when Fermi launches.

 

[toyota]

Originally posted by: BFG10K

Originally posted by: toyota

if thats true then how do you explain it losing to the 4870 and 4890? no matter how high you overclock the 5770 it cant touch a stock clocked 4890 in some games.

We can rule out the memory because it doesn?t seem to be the primary limitation based on my findings. There?s a bottleneck else where, and I?m convinced it?s at the driver level, intentional or not.

I don?t think we?ve seen the full performance of the 5xxx series yet, and I expect something interesting to happen when Fermi launches.

well it does seem odd that it gets raped by the 4890 so bad in that link I showed. in your Far Cry 2 results you have the 5770 beating even the gtx285 which seems odd. some other sites have the gtx260 not to mention 285 beating the 5770. http://techreport.com/articles.x/17747/6

 

[Qbah]

It would be relatively easy to determine. All we need is someone with both a HD4890 and a HD5770. Downclock the HD4890's memory to have same bandwidth and run the core at the same speed and compare the two

If the new ATi cards are intentionally crippled (for whatever bizarre reason), the HD5770 should still be slower than a downclocked HD4890.

 

[LCD123]

Originally posted by: toyota


if thats true then how do you explain it losing to the 4870 and 4890? no matter how high you overclock the 5770 it cant touch a stock clocked 4890 in some games. http://img360.imageshack.us/im.../1406/asushd577016.jpg

I posted a poll last month and alot of people actually thought hd5770 would match the hd4890. I explained that there was no way this was possible.

 

[ronnn]

Very nice review. My only quibble is that, who buys a 5770 to use at 2560 x 1600? Still some lower resolutions were used on occasion and it was all well thought out!

 

[toyota]

Originally posted by: ronnn
Very nice review. My only quibble is that, who buys a 5770 to use at 2560 x 1600? Still some lower resolutions were used on occasion and it was all well thought out!

well I think its good to show just how it scales and compares to other cards at that res.

 

[Ayah]

It's still a pretty terrible product for it's cost.

 

[thilanliyan]

Hey BFG, are you going to do a new IQ analysis with the 5770?

 

[TurtleCrusher]

Just a simple question, why did you use an E6850 on the review? Do you have any quad's or higher clocked duals?

No doubt a Core2 Duo @ 3ghz holds its own still, but it's just a question.

Edit: Woah it beats out a GTX 285 @ 1920 (Far Cry 2). Impressive.

 

[MarcVenice]

Originally posted by: Scholzpdx
Just a simple question, why did you use an E6850 on the review? Do you have any quad's or higher clocked duals?

No doubt a Core2 Duo @ 3ghz holds its own still, but it's just a question.

Edit: Woah it beats out a GTX 285 @ 1920 (Far Cry 2). Impressive.

That's weird. Must be caused by the fact 2xAA is used. In my own tests, a GTX 285 rapes a 5770, but that's with 4xAA. The GTX 285 scored 41,0min / 52,7avg and the 5770 scored 27,3min / 38,9 avg. Maybe it's a cpu thing too (we use core i7 @ 3,8GHz)

 

[Painman]

Seems to be a bit of a SKU without a cause... I dunno. It beats a 4850, so maybe it's in the neighborhood of a 4870, but doesn't really touch a GTX 260 overall.

That's a shame, because that would have made sort of a trifecta. 5870 within reach of dual GPU 295 and costing/consuming much less, 5850 beating up on 285, 5770 beating up on 260/275. But the 5770 falls short.

For the right here and now, 260 looks better. It trounces 5770 and can be had for a scant few $ more. Yeah, there's the promise of DX11. But that's not here and now. Budget card buyers might care about DX11, but who knows.

Anyway, props to BFG10K. I always appreciate the inclusion of older games into these reviews, but budget GPU buyers probably own budget monitors too. Cool it on the 2560 res benchies, OK?

 

[BFG10K]

Originally posted by: toyota

well it does seem odd that it gets raped by the 4890 so bad in that link I showed. in your Far Cry 2 results you have the 5770 beating even the gtx285 which seems odd. some other sites have the gtx260 not to mention 285 beating the 5770. http://techreport.com/articles.x/17747/6

I don?t think it?s odd because the 5770 currently has very inconsistent performance. In my tests you can see there are games where the 5770 is far faster than the 260+, but in others it?s much slower. Also in some games it?s barely faster than the 4850, yet in others it beats even a GTX285.

Like I said earlier, I don?t think we?re seeing the full performance of the 5xxx series yet.

As for Far Cry 2, it could be different benchmarks and/or different settings being used. I tested the 5770?s results repeatedly with multiple levels of AA, both in-game and through the driver, and everything was behaving as expected. I?m not saying the other benchmarks are wrong, but I?m confident the results for my particular settings are accurate too.

 

[BFG10K]

Originally posted by: ronnn
Very nice review. My only quibble is that, who buys a 5770 to use at 2560 x 1600? Still some lower resolutions were used on occasion and it was all well thought out!

Originally posted by: Painman
Cool it on the 2560 res benchies, OK?

Generally speaking, the settings I pick for each game are based on settings I?d use if gaming on one of the tested cards. That is to say, at least one of the tested cards must produce half decent scores at my chosen settings.

For example, the entire 1920x1200 page uses the settings I use on a GTX285 when I play those games, minus TrAA which I didn?t test. Sometimes there are small adjustments required to ensure highest common denominator settings (e.g. disabling TrAA/AAA), but most of the time they?re my actual gaming settings.

I do this because I don?t like throwing up wave after wave of theoretical settings, neither do I like putting up scores that are consistent slideshows, or scores that are so high they?re meaningless.

That and lowering the settings too far for the benefit of slower cards runs the risk of introducing CPU limitations for the faster cards. To me if I?m testing graphics cards, I should be straining them as much as possible while ensuring a playable score on at least one of them.

Additionally, by mixing up the settings I?m more likely to break drivers if they?re only optimized around common settings like 1920x1200 with 4xAA. This is also one of the reasons why I like benchmarking older titles too.

Anyway, that?s my benchmarking ideology.

These methods are somewhat unorthodox, but then I?ve never tried to create mainstream reviews. There are plenty of those around, so I always try to put up something a little different to the standard fare to get both myself and the readers thinking more.

Also even if the 5770 was my primary card, I?d still be using 2560x1600 in many games.

 

[BFG10K]

Originally posted by: Scholzpdx

Just a simple question, why did you use an E6850 on the review? Do you have any quad's or higher clocked duals?

No doubt a Core2 Duo @ 3ghz holds its own still, but it's just a question.

Originally posted by: MarcVenice

Maybe it's a cpu thing too (we use core i7 @ 3,8GHz)

I use an E6850 because that?s in my rig. The rig in my sig is my personal one, and also my benchmarking machine.

As for the CPU speed, I?ve demonstrated multiple times in the past why I believe CPU requirements are far overblown in online forums and review sites. This is especially true with high-end CPUs because the sorts of people running them will use high detail gaming settings, which largely eradicate differences between medium and high-end CPUs.

The way I benchmark, I can tell you the vast majority of my tests end up being massively GPU bound, and as long as the CPU is half decent (e.g. E6850), a faster CPU will make a negligible difference to most of the scores.

I demonstrated this by getting a ~30% performance gain from a GTX285 to a GTX260+, which is actually a lot higher than many mainstream sites showed, despite my CPU being much slower than theirs.

I?ll probably hold off upgrading my CPU until I get a Fermi so I can prove my point yet again. I can guarantee you there?ll be a massive performance gain over a GTX285 with my ?slow? E6850 @ stock, assuming Fermi doesn?t flop at the hardware and/or driver level of course.

Anyway, thanks for the great comments guys, and keep them coming, I really appreciate the constructive feedback, especially because it gives me a chance to explain my actions.

But most of all, thanks for talking the time actually read the stuff I put up. It?s readers like you that make the effort all worthwhile. :thumbsup:

 

[BFG10K]

Originally posted by: thilanliyan

Hey BFG, are you going to do a new IQ analysis with the 5770?

Yes, absolutely; I?m working on it right now. I?d argue that in many ways it?s actually more important than the performance test.

The IQ analysis is part 3 of my 5770 investigation, with part 1 being performance (linked to in the OP), and part 2 being the bottlenecking investigation. I envisaged three parts right from the start.

 

[frozentundra123456]

I am still not sure about the new ATI cards. For instance, the 5870 has about twice the processing power and only around a 50% increase in graphical performance. Seems like they have a bottleneck somewhere.

I usually root for ATI vs nVidia, but it seems that comparing the newer generation ATI cards to the old generation of nVidia cards just shows that if nVidia can get their act together and get their new cards out they could be ahead of ATI across the board.

It will probably ultimately come down to price vs performance. ATI may have to drop prices if the new nVidia cards are successful.

 

[nemesismk2]

Originally posted by: frozentundra123456
I am still not sure about the new ATI cards. For instance, the 5870 has about twice the processing power and only around a 50% increase in graphical performance. Seems like they have a bottleneck somewhere.

I usually root for ATI vs nVidia, but it seems that comparing the newer generation ATI cards to the old generation of nVidia cards just shows that if nVidia can get their act together and get their new cards out they could be ahead of ATI across the board.

It will probably ultimately come down to price vs performance. ATI may have to drop prices if the new nVidia cards are successful.

You can't blame ATI for there being no Nvidia video cards to do a proper comparison. I hope Fermi is as good as people seem to be expecting but what if it's another GeForce FX type video card? DX11 but too slow so have to use DX10 instead. I was gutted to have to play Half Life 2 in DX8 because my DX9 GeForce FX 5900 was too slow

 

[OmegaShadow]

yea... i'm also not sure about the new ati cards. im still using a 8800gs and am gonna wait for the new nvidia cards. i don't want to go through the trouble of changing drivers again.

 

[Pneumothorax]

Originally posted by: Ayah
It's still a pretty terrible product for it's cost.

Yup and unless you're paying dollars per Kw, a 4890 is a no-brainer. That's also why all the online stores have them in stock, unlike the 5850/5870. If it sold for lower than a 4890 and bridged the gap between 4890/5850 in performance, they'd be sold out too.

 

[Idontcare]

Originally posted by: Pneumothorax

Originally posted by: Ayah
It's still a pretty terrible product for it's cost.

Yup and unless you're paying dollars per Kw, a 4890 is a no-brainer. That's also why all the online stores have them in stock, unlike the 5850/5870. If it sold for lower than a 4890 and bridged the gap between 4890/5850 in performance, they'd be sold out too.

DX10 performance on 4890 is better but you get nothing for DX11 with that purchasing decision.

Personally I wouldn't be concerned with taking a 10% fps hit on DX10 titles in exchange for being able to play DX11 titles for the next year or so.