AMD Benchmarks Zacate APU, 2x Faster GPU Performance than Core i5

[god_43]

omg....we have turned this thread into a news article?...they link straight back to us...lol specifically kl0012. gj!



http://hothardware.com/News/AMD-Zaca...hmark-Hijinks/

[jimbo75]

omg....we have turned this thread into a news article?...they link straight back to us...lol specifically kl0012. gj!



http://hothardware.com/News/AMD-Zaca...hmark-Hijinks/

Yup I noticed that last night.

Btw looking at the kitguru video http://www.youtube.com/watch?v=pw14MgRHYJE (around 4 mins), you can see the window sizes actually go against the zacate machine. It's not much but the zacate system ran the browser tests at 1276x652 = 831952 pixels, and the i5 ran them at 1257x654 = 822078 pixels. If AMD was deliberately trying to hack the benchmark, they started off pretty badly.

[ajaidev]

[railer]

theres to much ricknrolin going on lately on these forums.

Need to see some game benches from APU. 18w is not much.

[Salvador]

http://www.anandtech.com/show/3933/a...ormance-update

Anand update.

[RaV[666]]

In this anand test, is the Core i5 running 2.4ghz with HT for the cpu part and Zacate 1.6ghz per core with no HT?
If thats true, than disparity gaming wise is HUGE, because i still believe bobcat cores are not that powerfull, athlon II levels AT MAX IPC wise.

[duploxxx]

http://www.anandtech.com/show/3933/a...ormance-update

Anand update.

so it was a i5 520M with a 766mhz GPU and zacate is about 50% faster in gpu performance, go find out how much slower the culv with 500mhz GPU is...

even with the SB mid next year for 18W parts and its possible GPU performance it will still be slower, not to mention the possible improvements from ontario in that time frame.

[AKM]

so it was a i5 520M with a 766mhz GPU and zacate is about 50% faster in gpu performance, go find out how much slower the culv with 500mhz GPU is...

even with the SB mid next year for 18W parts and its possible GPU performance it will still be slower, not to mention the possible improvements from ontario in that time frame.

I think that the gpu in Zacate is much faster than that 50% figure if you removed the huge disparity in CPU performance between that i5 cpu which turbos up to 3ghz and a 2 core 1.6ghz Bobcat.

[informal]

I think that Zacate may offer Turbo core functionality,but on more limited scale(think 2.2Ghz for single thread usage).

edit: Like i suggested some pages ago,it was intel GMA driver issue(laptop manufacturer's fault since on this model ,the official intel drivers won't install automatically).Anandtech confirmed it,thanks to the poster who provided the link for the update.

I borrowed a Core i5 notebook from Intel (we are at IDF after all) and ran the same tests on it. The performance was much better than what I saw in AMD’s suite.was ....

I brought up the issue and AMD immediately began looking into it. AMD used the latest drivers made available for this particular Core i5 notebook on the manufacturer’s website. Those drivers were version 2104 dated 3/31/10. AMD explained that these systems were configured about a month before launch and they were using the latest drivers available on the manufacturer’s website at the time. This particular OEM didn’t offer a newer version of the Intel HD Graphics drivers via its website. To make matters worse, Intel’s updated HD Graphics driver wouldn't install on the notebook. Some notebook OEMs prefer to provide their own display driver updates. In those cases, the Intel HD Graphics drivers available from intel.com won’t install. Despite only having a single Core i5 machine for all of its press demos for tomorrow, AMD agreed to let us manually install the driver (extract the driver and force device manager to update the driver manually) on the i5 system.

The updated driver brought the IE9 performance tests to parity with Zacate. In fact, it looks like the IE9 benchmark doesn’t scale too far with GPU performance (apparently discrete cards don’t score much higher than what we’ve seen here).

At this point we had an issue. The IE9 benchmarks AMD was showing off weren’t an accurate comparison of the two architectures. While valid for the only driver revision supported on this particular Core i5 notebook, the scores weren’t valid for a Zacate vs. Core i5 architecture comparison. AMD wanted to make sure there was no confusion about the GPU performance potential of Zacate so it allowed us to install whatever we wanted on both systems to validate the GPU performance we had seen.

Take a moment to realize exactly what just happened here. In an effort to convince us (and you) that it had nothing to hide and didn’t deliberately attempt to stack the deck, AMD gave us full access to the Zacate platform to do whatever we wanted. AMD wanted us to be completely comfortable with the Zacate comparison.

Summary:

Batman Zacate(16.5 fps) i5(11.3 fps)

Zacate is 46% faster. Note the disparity of CPU clock speeds on the Zacate and i5 : 1.6 Vs 2.4(2.933Ghz !).

City of Heroes : up to 2x advantage for Zacate,55% on average

N-Body Simulation : ~2.6x advantage for Zacate

AT says that Zacate is in no way optimized at this point in time(drivers,clocks,BIOS),there's still room for improvement.Batman ran for the first time on Zacate system... New driver didn't change the gaming performance of the i5,Zacate was still a lot faster(and up to 2x in City of Heroes ,like before).

So there is no "rigging" of tests,it was just the laptop with i5 that suffered from the GFX driver issue that is not uncommon.

[Hornet331]

so it was a i5 520M with a 766mhz GPU and zacate is about 50% faster in gpu performance, go find out how much slower the culv with 500mhz GPU is...

even with the SB mid next year for 18W parts and its possible GPU performance it will still be slower, not to mention the possible improvements from ontario in that time frame.

Think you missed this -> http://www.xtremesystems.org/forums/...&postcount=119

even with 200mhz clock it scores ~1700 points for the D2D test.

Also suddenly we saw a increse in CoH form sub 10fps to ~25fps on the intel platform...

Anyway we all know that intels current igps suck at gaming and a 80sp ati gpu will run circles around any current available igp. But for the most improtant part for laptops, 2D/HD content acceleration, both platforms offer the same results.

If you want to play games on a laptop get a discrete card.

[madcho]

zacate is just amazing !!!. I think CPU test are under NDA that's why anand don't published anything on.

Damm ... It's an amazing work, that was done by AMD.

[flyck]

Think you missed this -> http://www.xtremesystems.org/forums/...&postcount=119

even with 200mhz clock it scores ~1700 points for the D2D test.

Also suddenly we saw a increse in CoH form sub 10fps to ~25fps on the intel platform...

Anyway we all know that intels current igps suck at gaming and a 80sp ati gpu will run circles around any current available igp. But for the most improtant part for laptops, 2D/HD content acceleration, both platforms offer the same results.

If you want to play games on a laptop get a discrete card.

well it is targeted at low end/low power laptops and netbooks. These do not have the option to get an additional power consumer simply do to something useful in 3d. And we are still comparing to a 35W i5 cpu not the 18W with a much lower performance.. let alone the Atom series which are even slower than that.

[RaV[666]]

If you want to play games on a laptop get a discrete card.

Well obviously you live in a reality where everyone can afford high end gaming laptop with decent battery life.
But for majority of people thats not true.If entry level notebooks on zacate will be cheapish,smallish and with decent battery life, as it looks like it may be, and without Intel constant game driver issues, than it looks like a great mainstream product.
I often work as a kind of support ,and every other client is shocked that he cant game on his new intel notebook at all mostly.Sandy bridge may well change that, but until now Intel IGP suck ass.

[Bowtie]

So the CPU part is 32nm and the GPU part is 40nm? Or is the whole package 40nm?

[informal]

Whole package(monolithic) is 40nm bulk made by TSMC.

[Bowtie]

Whole package(monolithic) is 40nm bulk made by TSMC.

Cool, the first half node cpu ?

[mAJORD]

The Argument that you can't play games on that level of IGP is quite flawed.

You can play Quake III on an atom netbook.. but apart from nostalga reasons most wouldn't want to.

However there's plenty of current games that would play well on Zactate or even Ontario level of performance, that's the difference. It's perfect for those cheap indie specials on Steam for example.. which often have less than cutting edge graphics, but fresh and fun gameplay.

[Hornet331]

However there's plenty of current games that would play well on Zactate or even Ontario level of performance, that's the difference. It's perfect for those cheap indie specials on Steam for example.. which often have less than cutting edge graphics, but fresh and fun gameplay.

Most of this games run on any IGP that was released the last 4 years.

The only thing that would really bring igps forward is, if the next minumum standard would be LLano typ gpu in the lowest end cpu.

[flyck]

untill someone programs an application using openCL and Zacate blasts past most i models in computing power :P

[EvilOne]

Im looking foward to the day when i dont need to by a dedicated gpu, wont be long at the rate AMD are going. Games are already progressign alot slower the GPU speeds, so hopefully APU will catch up in the next few years. Results of this 'netbook' chip are prittey amazing for 18w, cant wait to see the better models.

[MrMojoZ]

"Take a moment to realize exactly what just happened here. In an effort to convince us (and you) that it had nothing to hide and didn’t deliberately attempt to stack the deck, AMD gave us full access to the Zacate platform to do whatever we wanted. AMD wanted us to be completely comfortable with the Zacate comparison."

I bet even after this you are still going to see "Waaah AMD lied!!!" posts for the next few months. 50% increase at this stage in actual games at that platform level is massive.

[mAJORD]

Most of this games run on any IGP that was released the last 4 years.

The only thing that would really bring igps forward is, if the next minumum standard would be LLano typ gpu in the lowest end cpu.

So there's only two types of games in your mind. Those you can play on an IGP from 4 yrs ago, and those you can play on a minimum of Llano class GPU?

[Manicdan]

I think that Zacate may offer Turbo core functionality,but on more limited scale(think 2.2Ghz for single thread usage).

i hope the future of turbo is able to do much more with fusion type chips.

think for a sec how a desktop has a 125W cpu, and 100W gpu. if you put them both into one package, you now have a 225W TDP, and you will then have a heatsink which is able to handle both. so when your gpu is not being used and you need better cpu performance, i think it would be incredible if the chip knew the load on both parts and adjusted turbo for both as needed. with some games only being duel core optimized, the chip should be able to shut off 2/4 cores, and turbo the cpu or gpu depending on which has a higher load.

the future of hybrid chips is really going to get interesting if we keep moving into that direction and get enough competition.

[ajaidev]

If AT had tested batman with Low Quality that would have been interesting and directly comparable to SNB preview results.

EDIT: Tried the N-Body Simulation on my friends desktop with a GTX 480 i am getting around 600-660 GFLOPS after which it crashes so i guess 23 GFLOPS is not bad for such a tiny little thing.

[-Boris-]

i hope the future of turbo is able to do much more with fusion type chips.

think for a sec how a desktop has a 125W cpu, and 100W gpu. if you put them both into one package, you now have a 225W TDP, and you will then have a heatsink which is able to handle both. so when your gpu is not being used and you need better cpu performance, i think it would be incredible if the chip knew the load on both parts and adjusted turbo for both as needed. with some games only being duel core optimized, the chip should be able to shut off 2/4 cores, and turbo the cpu or gpu depending on which has a higher load.

the future of hybrid chips is really going to get interesting if we keep moving into that direction and get enough competition.

I just want to add that I think it's reasonable to believe that you can save some amount of power when integrating parts. One chip instead of three different chips probably uses less power, and less power regulation is needed.

An interesting feature would be if the CPU throttled down to match the performance of the GPU. So if the GPU is capable of 60FPS and the CPU is capable of 130FPS, it could throttle down to save power without drop in framerate. Some of that power could be used to boost the GPU. A frame limiter at say 60FPS controlling the turbo and throttle would save even more.