Desktop Trinity benchmarks are up.

[Abwx]

It is an interesting observation. Can that much performance degredation be caused by scheduling and context switching and all of its associated overhead? If so then I'd say there is something seriously wrong with this architecture.

The dual core has more TDP headroom for a single or lightly threaded
task , hence the lower time thanks to higher turbo frequency.

 

[Kalessian]

Looks like to me the IPC is now on par with Thuban with the module penalty kicking in with multiple threads.

Pretty good assessment, yes sadly i can foresee oc'ed vishera might still lose to oc'ed x6 thuban in heavy fp workloads that are not avx-aware.

 

[Haserath]

Well looking at this graph we can get a little more insight into ph2 vs piledriver (seems the wav file is a little shorter in this ph2 test, also the ph2 runs at 3.7ghz, not 3.8ghz and itunes version isn't the same so it's far from perfect):

The 8150 is probaly turboing to 3.9ghz so call it 3.8ghz average? Wav seems to be ~5% longer, but ph2 980 is ~2% slower [than 3.8ghz]... I scaled the ph2 at 3.8ghz to be 1:25 compared to 1:31 for trinity Really dirty comparison but trinity looks to have just about caught up with ph2. Is llano better than ph2 clock for clock?

How about instead we look at these two:sneaky::
http://www.tomshardware.com/reviews/a10-5800k-a8-5600k-a6-5400k,3224-13.html

http://www.tomshardware.com/reviews/fx-8150-zambezi-bulldozer-990fx,3043-17.html

It's really weird that the 5400k can only turbo up to the base clock of the 5800k, but the 5400k still wins.

 

[ttechf]

I was waiting for benches like this to come out because I'm wondering if I should go AMD Trinity or Intel i5 2500k for my first system build? : )

 

[pelov]

Getting mighty close to high settings playable at 1080p with no discrete card.

1 more generation and I think we might see the end of all GPUs under $100 for good. 1080p at 30FPS + and high settings? Yesplz

I think I might be close to tossing my desktop in favor of a small APU HTPC. It looks like I may get my wish in 2013. Couple that with a nice Asus tablet and dock and I'm all set

 

[Homeles]

Where are you guys getting IPC increase of 15%. Certainly not based on the benchmarks in this preview. OR do we do % differant for AMD than intel . If trinity is a 15% increase in IPC intel IVB is 25% faster IPC than SB. Funny you AMD guys are . Alot like the Dems.

From this:

Keep watching your Faux News, kid. And your math is atrocious.

 

[Kalessian]

How about instead we look at these two:sneaky::

Cool, thanks. My 1:25 was close

Edit: Yeah it is weird that the 5400k does 1:25 yet in the 3.8 no turbo test trinity only does 1:31...

 

[KompuKare]

Toms really fail's at benchmark reviews.

+1. The irony is the reviewer did leave this reply in the comments:

"Don't worry--I'm working on the data right now. As it stood, this story took more than a week of all day/all night testing, troubleshooting, new BIOS installing, and re-testing to nail down. It can go on indefinitely if you let it ;-)"

Ok, previewing pre-lease hardware with flaky BIOSes cannot be nice but don't Tom's have a database of tests they keep so they can at least include some other CPUs or does each reviewer start from stratch. And that power usage chart is interesting but totally unreadable...

 

[ShintaiDK]

It is an interesting observation. Can that much performance degredation be caused by scheduling and context switching and all of its associated overhead? If so then I'd say there is something seriously wrong with this architecture. It is already hugely disappointing that it just barely beats llano even though it is running 30% faster.

I dont see how this competes with a llano 3870k when both are overclocked. The trinity is very likely to top out at around 4.4GHz, which wont beat a 3.3GHz llano much less a 3.6GHz llano.... (gaming aside)

Yep, remember the Bulldozer hotfixes? Looks to be the same case essentially. Less cores, less jumping. Also why the 5400K wins.

A new run with all but one module disabled on the FX8150, 5800K, 5600K and 5400K would be interesting to test if thats the case.

If Toms had done any multithreaded tests we might have seen it too.

 

[grimpr]

Pretty good assessment, yes sadly i can foresee oc'ed vishera might still lose to oc'ed x6 thuban in heavy fp workloads that are not avx-aware.

Remember, an 8 threaded BD/PD has 4 Floating Point Units compared to Thubans 6, an 4 threaded Trinity A10 has 2 Floating Point Units compared to Llanos 4.

 

[Nemesis 1]

From this:

Keep watching your Faux News, kid. And your math is atrocious.

Ya kid. So you say 15% based on 1 benchmark. Its how we do it for intel is we take the average after throwing the high and lows out . So when AT does his review based only on the same programs that AT did in the IVB review and removing the high and low than your average is what AT . Thats at the same clock as last generation . or its not just IPC members call the ipc increase.

 

[pelov]

Yep, remember the Bulldozer hotfixes? Looks to be the same case essentially. Less cores, less jumping. Also why the 5400K wins.

A new run with all but one module disabled on the FX8150, 5800K, 5600K and 5400K would be interesting to test if thats the case.

If Toms had done any multithreaded tests we might have seen it too.

Yea... that's stupid. It's the most disappointing part of the CMT approach, I think. And nobody should be forced into buying a crappy operating system, win8, in order to reap the full benefits of the new scheduler that handles their architecture better.

Or you can go Linux

 

[Homeles]

Ya kid. So you say 15% based on 1 benchmark. Its how we do it for intel is we take the average after throwing the high and lows out . So when AT does his review based only on the same programs that AT did in the IVB review and removing the high and low than your average is what AT members call the ipc increase.

When you have acceptable English and grammatical skills, you can attempt to have an argument with me. Until then, you are not remotely worth any of my time.

Yep, remember the Bulldozer hotfixes? Looks to be the same case essentially. Less cores, less jumping. Also why the 5400K wins.

A new run with all but one module disabled on the FX8150, 5800K, 5600K and 5400K would be interesting to test if thats the case.

If Toms had done any multithreaded tests we might have seen it too.

Yeah, it looks like it's an OS scheduling issue, not an inherent flaw of the architecture.

 

[Riek]

Yep, remember the Bulldozer hotfixes? Looks to be the same case essentially. Less cores, less jumping. Also why the 5400K wins.

A new run with all but one module disabled on the FX8150, 5800K, 5600K and 5400K would be interesting to test if thats the case.

If Toms had done any multithreaded tests we might have seen it too.

This could indeed be the case. But those fixes didn't really bring any noticeable performance gains, so nothing he can do about it.

edit: i thought the difference was ok, but its rather big with close to 9%. Curious if windows 8 really improves for this by that much.

 

[Nemesis 1]

Debating AMD fanbois is a waste of time . But I have time to waste Mister nazi grammer cop. Trinity sucks as a CPU. It does GPU things better but not good enough . When its good enough I will never buy another GPU . HASWELL! You have a writing form I have seen befor . This is the second time.

 

[Homeles]

Debating AMD fanbois is a waste of time . But I have time to waste Mister nazi grammer cop. Trinity sucks as a CPU. It does GPU things better but not good enough . When its good enough I will never buy another GPU . HASWELL! You have a writing form I have seen befor . This is the second time.

Your English skills are deplorable, and I should not have to spend 10 minutes translating the piles of garbage that your spew. If you're going to try having a discussion on a forum, I strongly suggest you learn how to communicate your thoughts effectively.

Also, where's your proof that I'm a fanboy?

 

[sm625]

Yeah, now has similar IPC as a 2007 intel CPU...great achievement.

I dont even see that. I'm sure a Q6600 at 3.8GHz would lay waste to these chips at the same clock. Even at 3.3GHz, a Q6600 should be faster.

 

[Homeles]

Yeah, now has similar IPC as a 2007 intel CPU...great achievement.

With that logic, a processor that can handle 1 million instructions per clock (which is ludicrously high) that refreshes at 1Hz is faster than a 1 instruction per clock processor refreshing at 3 billion Hz.

IPC is completely irrelevant without taking clock speed into account. Instructions per second is the metric that matters.

 

[Nemesis 1]

I don't need proof My animal instincts are strong . The sent of AMD is all over you. Nice attempt at diversion to the facts . Grammar nazi! You want to debate english skills I suggest you go to a british forum and debate your english skills with theirs. To declare IPC increase for Trinity at this point is a no go. We need AT review. Run the exact same benchmarks that was used for IVB review from there we can get a feel for trinities true compute performance. I don't care What CPUs people buy or use . But the constant lieing by AMD fanbois is wearing thin . 6 years and counting now . Nothing but lies.

The constant trolling by you is also wearing thin. So knock it off
-ViRGE

 

[inf64]

@ veri745
You are correct.

It is an interesting observation. Can that much performance degredation be caused by scheduling and context switching and all of its associated overhead? If so then I'd say there is something seriously wrong with this architecture. It is already hugely disappointing that it just barely beats llano even though it is running 30% faster.

I dont see how this competes with a llano 3870k when both are overclocked. The trinity is very likely to top out at around 4.4GHz, which wont beat a 3.3GHz llano much less a 3.6GHz llano.... (gaming aside)

Are you looking at the same preview as I am?

Sandra 2012: 1)Arithmetic : ~equal in Gflops and Trinity 24% faster in Gips;2) Multimedia: Float.mpix/s Trinity is 2.13x faster , Int.mpix/s Trinity is 2.15x faster;3)Cryptography: Trinity is 5.36x faster in AES256 and 1.3x faster in SHA256
Adobe Photoshop CS5: Trinity 5800K is 23% faster
Adobe Premiere Pro CS5.5: Trinity 5800K is 13% faster
Adobe AF CS6: Trinity 5800K is 17.5% faster
3dsmax 2012: Llano is 2.7% faster (within margin of error) <-fp heavy workload ,4 Llano fp units vs 2 PD FP units
Solidworks 2010 : Llano is 1.8% faster (within margin of error) <-fp heavy workload ,4 Llano fp units vs 2 PD FP units
ABBY Finereader 10: Trinity 5800K is 12% faster
Fritz: Llano is 3.4% faster <-fp heavy workload ,4 Llano fp units vs 2 PD FP units
Visual Studio 2010: Llano is 4.4% faster
Adobe Acrobat X: Trinity 5800K is 55% faster
Mainconcept: Trinity 5800K is 10% faster
Handbrake: Trinity 5800K is 6.2% faster
Lame: Trinity 5800K is 21% faster
iTunes: Trinity 5800K is 18% faster
Winzip: Trinity 5800K is 15% faster
Winrar: Trinity 5800K is 20% faster
7-zip: Trinity 5800K is 7% faster
OpenCL: Winzip(equal),Luxmark(Trinity 5800K is 22% faster)

Average adv. for Trinity 5800K (versus 3850 @2.9Ghz): 15% without Sisoft sandra and ~22% with Sisoftsandra results. Without Fritz and sisoft sandra,so only real world workloads, 5800K is ~16% faster than 3850 @ 2.9Ghz.
As for OCing,I googled a bit and most reviewers got results between 3.6 and 3.8Ghz from their 3870K samples with 3.8Ghz having stability issues in most cases(being barely stable with air cooling methods). So it's safe to assume 3.7Ghz for 3870K is practical limit for air cooling.
On the other hand,with this early platform THG managed to get 4.5Ghz out of Tirnity while AMD claims that with their (more mature/stable?) system they have managed 4.8Ghz,which IMO is very likely to end up being the limit for air cooling. So ~4.8Ghz vs 3.7Ghz on air cooling. 4.8Ghz is around 20% faster than stock 5800K which usually runs at 4Ghz due to Turbo. 3.7Ghz on Oced Llano is 27% faster than 3850. So in the end,the OCed 5800K @ 4.8Ghz should be around 10% faster (according to THG numbers above) than 3870K @ 3.7Ghz. Llano on average has 17% higher IPC than Trinity and 6% higher IPC than Phenom II. In turn, Phenom II has ~11% higher IPC than Piledriver and 20% higher IPC than Bulldozer. Piledriver has around 10-15% (depending on benchmark) higher IPC than Bulldozer,or closer to ~10% on average.

To summarize,5800K @ max OC (~4.8Ghz estimated;on par with FX4100's OC results ) is still better option than 3870K @ max OC (3.7Ghz based on online reviews with air cooling). Difference is around 10% in CPU tests. GPU portion is also overclockable in both CPUs and one is to expect that Trinity will hold its lead or extend it to more than 30% with max. OC on both CPUs.

 

[Homeles]

I don't need proof My animal instincts are strong . The sent of AMD is all over you. Nice attempt at diversion to the facts . Grammar nazi! You want to debate english skills I suggest you go to a british forum and debate your english skills with theirs. To declare IPC increase for Trinity at this point is a no go. We need AT review. Run the exact same benchmarks that was used for IVB review from there we can get a feel for trinities true compute performance. I don't care What CPUs people buy or use . But the constant lieing by AMD fanbois is wearing thin . 6 years and counting now . Nothing but lies.

What can be asserted without evidence can be dismissed without evidence, Since you refuse to provide proof that I'm an AMD fanboy, your assertion can be dismissed.

The funny thing is, I agree with you. We do need a more thorough review. I was just destroying your ridiculous claim that there wasn't anything showing IPC increases in this review, when Tom's specifically addressed that issue.

There's no lies here. Just just ignorance -- and loads of coming from you.

 

[Soulkeeper]

my 3870K at 3.4GHz generally smashes FX4100's at 4.4GHz in benchmarks

Here is an example of my system laying waste to one here

even with a 15% increase in performance that system would still lose 2/3 of those benchmarks at 4.66GHz

 

[Magic Carpet]

My first impression from this preview is a bitter disappointment. Still relies on high clock to be relevant. Mini-Netburst, all over again. But let's wait for the retail samples first

 

[Homeles]

Still relies on high clock to be relevant. Mini-Netburst, all over again. But let's wait for the retail samples first

If the end result (high clock + low IPC) ends up being faster, what's the problem with it? We're not even a year into the Bulldozer era from AMD... it's too early to chalk it up to being a P4 disaster.

 

[Nemesis 1]

Well The fact that you said I said there was no IPC increase is amusing . No where did I make such a claim . . But if your going to Figure IPC . Its has to be the same method that is used to determine Intel IPC increases. Clocks normalized and run the benchmarks ./ Its that simple . IVB was run against the same clocked SB . normalized. A 15%IPC increase for trinity is = to about a 10% increase in IVB.