AMD Piledriver rumours ... and expert conjecture

[Reynod]

We have had several requests for a sticky on AMD's yet to be released Piledriver architecture ... so here it is.

I want to make a few things clear though.

Post a question relevant to the topic, or information about the topic, or it will be deleted.

Post any negative personal comments about another user ... and they will be deleted.

Post flame baiting comments about the blue, red and green team and they will be deleted.

Enjoy ...

 

[palladin9479]

Anandtech got 2.05 for Cinebench multithreaded, a .04 variance which could be due to any number of factors.

I also want to see what THG gets, but do they actually use Cinebench? Not sure they do since their IB mobile benchmark didn't have it.

They got 1.88 on a 3500M and 2.04 for a 3520MX, I'm doubting them already. Plus it's Anandtech, they could show a P3 beating a BD if they wanted to. I take everything with a grain of salt from that place.

I wonder if anyone checked to see if K10 stat worked with it and tested it on that level, would show a clearer picture. Go back a few pages to see what I could get with just a few tweaks, especially with the processor affinity flag.

To clarify, 2.0 is a bad number to get for anything with four cores regardless of their arrangement.

 

[jimmysmitty]

Looking over ... I think they played with the turbo boost more then anything. If TB was disabled on the AMD chip then it would of been stuck at 2.3 regardless, enabled and the Intel chips would of hit 2.8 and 3.9 respectively. They mentioned down-clocking but didn't say anything on how the TB was set to.

Also 1.99 multicore (they didn't list single core nor the multiplier) on CB11.5 is pathetic. The 3530MX firing at 1.9Ghz x 4 gets higher then that, even at 1.7 it gets better then that. The 3550MX I've pushed to 2.7 x 4 and got really nice numbers, although I doubt it could sustain 2.7 x 4 with the GPU firing on full. I know it can sustain 3.0Ghz on a single core with the other three at 800 and the GPU firing on full.

Also no way in hell the HD4 got near the APU's graphics unless something very very bad was going in with the system memory. I'm also suspecting foul play there, BD's IMC was bad, but that that bad, their saying Trinity is worse then BD basically.

Trinity is BD based but mobile. IB mobiles memory bandwidth is lower than IB desktops memory bandwidth (DT gets about 20-22GB/s with DDR3 1600, mobile hits about 15GB/s) so it would not suprise me that Trinities mobile version based on the same base arch as BD would have worse memory bandwidth.

As for the APU, why is it so suprising? THG had a benchmark of the HD4K vs Llanos top mobile part and it was much better overall. The mobile side is not going to be nearly as fast as the desktop side, thats a known. But I would like to see your numbers for your Llano CPUs. But for Intel, HD4K is still HD4K. The mobile and desktop parts will be the same, only the CPU speeds and performance will vary.

Again, remember this is BD based while Llano was Stars based (K10.5 Athlon II series). Llano has 4 actual cores, trinity is using 2 modules which make up to 180% of 4 cores, not 4 full cores, which can affect performance in Cinebench.

Still, two site has nearly the same scores in Cinebench. And again I don't think THG will have Cinebench but we can at least compare other benchmarks like PCMark 7 and a few games. We should see the THG review in a few hours if not tomorrow early morning.

 

[triny]

More Trinity Benchmarks! ~7% better IPC than Phenom II, ~43% better single core performance than Llano! (K 10.5)

http://amdfx.blogspot.ca/2012/05/more-trinity-benchmarks-43-better.html

 

[noob2222]

80% of a dual core, not 100%+ another 80%.

The vr zone article was a bit odd. Trinity vs intel only.

I want to see some downclocked 4100 comparisons to see how they pair up even with the lack of l3

 

[palladin9479]

Trinity is BD based but mobile. IB mobiles memory bandwidth is lower than IB desktops memory bandwidth (DT gets about 20-22GB/s with DDR3 1600, mobile hits about 15GB/s) so it would not suprise me that Trinities mobile version based on the same base arch as BD would have worse memory bandwidth.

As for the APU, why is it so suprising? THG had a benchmark of the HD4K vs Llanos top mobile part and it was much better overall. The mobile side is not going to be nearly as fast as the desktop side, thats a known. But I would like to see your numbers for your Llano CPUs. But for Intel, HD4K is still HD4K. The mobile and desktop parts will be the same, only the CPU speeds and performance will vary.

Again, remember this is BD based while Llano was Stars based (K10.5 Athlon II series). Llano has 4 actual cores, trinity is using 2 modules which make up to 180% of 4 cores, not 4 full cores, which can affect performance in Cinebench.

Still, two site has nearly the same scores in Cinebench. And again I don't think THG will have Cinebench but we can at least compare other benchmarks like PCMark 7 and a few games. We should see the THG review in a few hours if not tomorrow early morning.

Trinity, like Llano, is the exact same chip for desktop (Llano) vs laptop (Sabine) just clocked lower and with various features enabled / disabled along with a different socket (FM1 vs FS1). That is why I was able to get it to run so fast, my only limitation was the notebook cooling solution. Due to this it's memory performance should be identical or near identical. Now I haven't seen a Trinity desktop / laptop CPU side by side thus I can't compare them, but I would assume (dangerous I know) that it would be the same die on both desktop and laptop models. This is another reason I'm concerned, those numbers would make perfect sense with a 2.0Ghz clocked BD CPU, but BD clocks higher then K10 hence its 2.3 ~ 3(3.2 not sure) range for TB. Trinity should be rocking 2.8Ghz at least seeing as I can get my K10 to 2.7 x 4.

I expected cache performance to be bad, I didn't expect half the memory bandwidth (which would explain the poor GPU results).

And you can't say "two sites", one is anandtech which started their Sabine review trying to sell you on a SB and I wasn't kidding on expecting them to show P3 beating a BD. Hopefully when Toms does their they will investigate on any deficiencies and ask the manufacturer about it, like they've done in the past.

 

[palladin9479]

80% of a dual core, not 100%+ another 80%.

The vr zone article was a bit odd. Trinity vs intel only.

I want to see some downclocked 4100 comparisons to see how they pair up even with the lack of l3

I've always treated BD cores as "0.75" of a full blown core, mostly due to L2 cache and instruction decoder arbitration issues. They mentioned tweaks and fixes to those for Trinity (BD-E) so I'll have to re-evaluate that number. You can tell whats really going on by playing with the affinity bit in Windows.

 

[sarinaide]

More Trinity Benchmarks! ~7% better IPC than Phenom II, ~43% better single core performance than Llano! (K 10.5)

http://amdfx.blogspot.ca/2012/05/more-trinity-benchmarks-43-better.html

I would think that most of us AMD users would hope that is correct, the numbers coming out of late seem encouraging compared to numbers mentioned in the lull period, time will tell. If the IPC numbers are as suggested 10-20% faster (somewhere in between) that would be a very reasonable improvement, maybe not to match Intel but AMD essentially competes against its own previous generation chips, so that is a major improvement.

 

[amdfangirl]

I hope for two things:
*Good Performance/watt for Piledriver
*Gigabyte to make my motherboard support Piledriver

😛

 

[sarinaide]

Zambezis stock power consumption is still higher than a 2600K from Intel and if you knew AFG, you would know she likes low power. If anything she would get the lowest TDP CPU then undervolt the crap out of it to have an even lower power consuption CPU.

Temps are one thing, power consompution is another. A 2600K will stay at under 50c when running normal loads (mine does about 55-60c but thats overclocked to 4.5GHz) and idle around 30-35c. BD is about the same at stock and still doesn't perform as well. Well lower would be 10-20c lower under load.

The major power consumption issue with AMD is that most enthusiasts like to overclock, and AMD has been marketing overclocking to the more budget oriented enthusiasts. Well, BD has major jumps in power usage. At 4.8GHz it uses more power than a 2600K at 5GHz on the same process size. Then again, Intel is on their second gen HK/MG as well as has had time to mature 32nm while AMD has no hand in it and cannot control it.

I did gather that AFG is a eco friendly builder from all of the stickied threads

Yeah I would expect a 8 core processor to use more power than a quad, but for me power consumption is a less important factor, and when it is not significant it becomes even less redundant, but we are all to our own.

AMD have had to make do in the best manner possible within the financial limitations of the company, considering Intel monopolise just about every aspect of the DT industry and Mobile industry along with the superior facilities it is difficult to fathome AMD ever beating Intel, nor are any AMD users here under the misguided impression that PD/Trinity is going to change that, but I thank you for keeping us honest nonetheless. 😀

 

[esrever]

trinity is out. Seems pretty decent.

 

[Haserath]

Wow, just looked at anandtech and saw the Trinity article. I must say I'm pleasantly surprised by it as well. Better than I thought it would be.

 

[noob2222]

found a semi-comparison with current BD cpus.

http://www.cpubenchmark.net/cpu_lookup.php?cpu=AMD+A8-4500M+APU

not exactly the best place to find a comparison, but dang, 1.9 ghz 2.8 turbo keeping heels with the 4170 4.2/4.3 ghz ... who knows what the actual speed tested was.

It is however a bit above the a8 3600 2.1/2.4 ghz cpu. only thing is, does that take into consideration the gpu? Could explain the difference to the 4170 unless its just that much better. I can see 20%, but near 50 is hard to believe.

Like i said, not reliable but at least there is a hint for what PD may actually be bringing in.

Most of the reviews I have seen so far just pit Trinity up against 2-3 intel chips, most don't even show Llano ...

One thing is very clearly impressive tho, Battery life ..

 

[sarinaide]

Wow, just looked at anandtech and saw the Trinity article. I must say I'm pleasantly surprised by it as well. Better than I thought it would be.

Well it depends on the context and light you look at it, or are you convinced that Intel and AMD are playing at the same game.

 

[Haserath]

Well it depends on the context and light you look at it, or are you convinced that Intel and AMD are playing at the same game.

I'm looking at it from the POV of 'they improved over their last design pretty well.' If you compare it to Intel in CPU performance, of course it loses, but Trinity improves on Llano's performance and increases battery life in the process. Better than they did last time. 😉

If it actually competed with Ivy Bridge in CPU and beat it in graphics, that would have been 'high on <insert whatever of choice here>' instead of just pleasantly surprised.

Haswell may change Intel's standings in graphics. We'll see.

 

[sarinaide]

At this point we don't have DT numbers to go by, but it is overall more promising than maybe 2-3 months ago which was rather doom and gloom. AMD competes with its prior generation and it seems that Piledriver a.k.a Bulldozer part deux is a step in the right direction, if the numbers pouted are anything in that 10-20% window across the board then it is a definate worthy upgrade from the ol hammer that is presently sitting pretty in my rig.

And again, the days of AMD competing with Intel are gone, money and resources wise AMD cannot go toe to toe with Intel, it is more a case of punching smart like Ali, rather than going in fists flailing Tyson style. There is nothing wrong with AMD's architecture, it just needs maturity.

 

[-Fran-]

I'm also quite surprised with the battery life Trinity pulled out. Llano has terrific battery life, but Trinity does even better in the same process with more transistors. That's quite a feat from AMD's engies.

Also, notice in the performance ballpark where Trinity lands... It's right in the doorstep of USD800+ Intel notebooks. If Trinity comes under USD700 that's going to be quite sweet across the board for notebooks and ultra-thins/portables.

Also, the IPC increase is not that noticeable, but you can tell it's there. They're not clocked that far from each other, yet Trinity is actually pulling ahead. Maybe not by a wide margin, but when you think it's using all the TDP it can against Llano, there should be like a 300-400Mhz difference only, which could put them in the same IPC ballpark.

IMO, TH's review will put some more perspective for all of us, but so far so good.

Cheers!

 

[truegenius]

*Gigabyte to make my motherboard support Piledriver

+1
since my board is giga one 😀
with
am3+ socket
black socket
32nm ready
4 phase cpu power design
am3+ ready
125 tdp support
880g chipset

viola, smells like pd

AMD cannot go toe to toe with Intel, it is more a case of punching smart like Ali, rather than going in fists flailing Tyson style.

+1
thats AMD 😛

 

[jimmysmitty]

I'm also quite surprised with the battery life Trinity pulled out. Llano has terrific battery life, but Trinity does even better in the same process with more transistors. That's quite a feat from AMD's engies.

Also, notice in the performance ballpark where Trinity lands... It's right in the doorstep of USD800+ Intel notebooks. If Trinity comes under USD700 that's going to be quite sweet across the board for notebooks and ultra-thins/portables.

Also, the IPC increase is not that noticeable, but you can tell it's there. They're not clocked that far from each other, yet Trinity is actually pulling ahead. Maybe not by a wide margin, but when you think it's using all the TDP it can against Llano, there should be like a 300-400Mhz difference only, which could put them in the same IPC ballpark.

IMO, TH's review will put some more perspective for all of us, but so far so good.

Cheers!

The battery life could be many factors such as larger battery or possibly better power management features which I would expect out of Trinity.

But one thing is I am a bit dissapointed THG didn't throw HD4K in there for good measure. I am sure Intel will have one in the price range so its not a bad thing to at least test it now to see what we are looking for. Everyone else is testing HD4K against Trinity/Llano. We have to consider that a bit down the road Intel started to throw out revisions of the lower end CPUs without HD3000 that had HD3000, so why not just test HD400 now and get it out of the way?

Just seems like it was a bit strange not to include it because some people wont be driven by price but rather performance overall in certain areas.

Trinity is very much better than Llano so it seems. I still want to see per clock and per core overall performance but thats just me.

The memory bandwidth in PCMark 7 looks the same as with AIDA64. COuld be due to a new CPU but it should not have issues with Llano. Also, laptops right now stick around 1333 DDR3, 1600 DDR3 looks to becoming normal but I haven't seen many with it yet.

 

[fazers_on_stun]

The one you linked shows 20-25% in favor of IB, using a dual core with SMT clocked lower (2.4GHz vs 2.3GHz with 3.8GHz being max and top end IB mobile is 2.9Ghz quad core with turbo to 3.8GHz). Anandtech showed overall the CPU is still weaker than IB.

The GPU was not what I expected at all. I expected it to plow HD4K, not just slide by (IB did win some games by a decent amount).

I expected Trinity to be much stronger in the GPU side due to AMDs advantage of doing GPUs much longer. But its not.

Yeah, the 15-game average showed the 7660G about 20% faster than the HD4K, about 19% faster than the 6620G (Llano igp) and a huge 80% faster than the HD3K on Sandy...

 

[JAYDEEJOHN]

Going to GCN next, making a few more changes allows AMD some wriggle room down the road, as their power usage looks very good.
Its a bit better than what theyd originally said overall, which is also good to see, and to jimmy, read what cleve said in the comments as to why no IVB

 

[truegenius]

and mal~ rocks a "GA-990XA-UD3" so what about it.?
can I sell you mine.?

if it is completely free then let me have it 😀

if it isn't then
😗 oh where is my piggy bank
😱 my thuban ate that piggy 😗
also i5 is waiting to be with me 😀

 

[fazers_on_stun]

Haven't checked many other reviews that AT & THG's and a couple linked in this thread, but a few early observations:

1. FMA4 doesn't seem to be the incredible performance booster that Baron was stating last year, and others since then.

2. Although Llano boosted AMD's mobile marketshare some 2.6% last year (think AMD is up to 16% now, according to IDC and contrary to Triny's wild predictions 😛), I dunno if Trinity will continue to push that trend, esp if it is not widely available by June when the IB laptops supposedly will be. If it is late Q3, AMD will miss the back-to-school bump in sales. And to use the AMD fans own argument against them, it seems IB's igp is "good enough" anyway - it'll all boil down to how competitive pricing will be IMO, for those budget gaming laptops.

3. Improving CPUs or APUs is pretty hard. A lot of Intel fans were disappointed that IB was not the killer CPU that they were hoping for. Most of us were expecting 10-20% performance gains (but then that was early, and also in comparision to the 2600K, not the 2700K that was used in most of the reviews last month). And now Trinity doesn't appear to be quite the killer CPU that most of the AMD fans were hoping for, at least in the ones benched in today's reviews..

 

[JAYDEEJOHN]

I disagree, the cpu is quite nice in improvements.
Those lappys are coming, so they wont miss, DT however is another matter.
Im not sure who said they expected alot more out of the cpu, but I havnt seen it, except from the usual suspects

 

[sarinaide]

On the mobile perspective the Trinity APU's are marked improvements on the weak Llano's, it more than competes well with the Intel mobile equivalents and that is on the mobile platform.

This is very encouraging for the DT models to come.

 

[jdwii]

I disagree, the cpu is quite nice in improvements.
Those lappys are coming, so they wont miss, DT however is another matter.
Im not sure who said they expected alot more out of the cpu, but I havnt seen it, except from the usual suspects

Like i said i'm quite proud of those improvements The A10 should be priced at Low-end I5's and I3's(600-680$). At this price this APU will be a killer but anything higher then that it will not be a killer.

If amd can make me a 4.2Ghz 8 core Piledriver with L3 and price it at the I5 i'm in! I see a 7% improvement in IPC(on Average compared to BD) and a 43% increase in CPU frequency(Compared to the A8 3500M) that made the APU 20% faster overall compared to the A8 3500M in CPU tasks.

I agree with noob2222 i would like to see a 4 core BD clock down to 2.3Ghz and compared to this Chip. That should give us a better clue on improvements for Piledriver. Because unlike Llano-Trinity BD-PD can't be clock 43% higher which would leave a smaller improvement. Even if they can clock their CPU to 4.2Ghz its only 17% higher then the 8150. Let me try to take a guess of PD improvements if they can clock their CPU to 4.2Ghz.

Trinity to PD will probably be 7-10% faster per clock on average because of L3 cache. Trinity looks to have 7% better IPC compared to Bulldozer. The CPU will also probably be clock at least 10% higher.

So i expect the new 8 core PD to be around 15% faster then BD at launch with the possibility of 20%(all based on average performance not "upto" statements) but nothing greater. But its also probably going to be 10-20% more efficient per watt as well.

So to sum it up that's my last guess based on today's reviews and yesterdays news.
BD-PD will be 15-20% faster overall and 10-20% more efficient.

Does anyone else have any estimates i would love to here them! Try to be realistic.