AMD "Piledriver" refresh of Zambezi - info, speculations, test, fans

[informal]

Yes,in pure numbers it's around 26% (actual BD is 26% slower than what I anticipated in the chart). Most of it is due to lower IPC in fp/simd workloads.But Piledriver has at least a chance to remedy some of that. 5% on core improvements and a 4-4.1Ghz base clock and we are at 210 "pts" in that chart for Piledriver. Not bad but not as good as I hoped either. But that's life,things just don't turn out as good as we hoped.

[undone]

Severely off-topic, guys. If you still wanna argue about bdver1 pls go to other thread or submit new thread, this thread is for trinity. I'm frustrated now.

source

thx for the info. but B3 Zambezi is still bdver1.

Of course we can expect a better Trinity according to this info.

[Zeus]

I do understand people cry about bulldozer. Look at the facts: it took AMD what, 4 years to develope this new uarch? After a 4 year wait expectations are sky high. You'd expect a product that has been tried, tested, optimized and enhanced as much as possible. You'd expect a product that can compete with Intel and something that is at least a whole lot faster than it's predecessor.

What do you get? A cpu that runs blistering hot, comsumes more power than all the rest of the electrical devices in and around the house together and yet has an IPC that is lower than it's predecessor and not just by a small margin.

How can you speak of a quality product after all that? Overall faster than Thuban? Someone forgot to give me that memo?

Isn't it kinda rediculous we have to wait another timespan (god knows how long) for Piledriver to get what Bulldozer was supposed to be?

Wake up and smell the coffee.

[wez]

Elaborate.
You must be even more 'full of it', you insult other members as if you are better than everyone.
Back yourself up, explain how I'm so 'full of it'.

Look over your posts and then tell me your not pulling stuff out of thin air. If not, then show us the "facts" you so boldly claimed?

[-Boris-]

Seriously, I have no idea why people are still crying about Bulldozer. Sure it's not the fastest chip,but at least it is overall faster than Thuban(sometimes a lot ,sometimes it's even slower). The good thing is that even in this worst case scenario when both design and process have some issues,it's quite a decent product. It just needs a price correction to be a bit more competitive from price/perf. POV.
Piledriver will improve on Bulldozer,Steamroller on Piledriver etc. The curve goes to the right direction so I don't know why some people are so pessimistic... If anyone thinks that AMD could have rode K10 on sub 32nm with 4+GHz clocks and 8+ cores with no changes to the uarchitecture itself is delusional.

Because we are not too fond of turd polishing! It’s a botched architecture, you can’t just give it 50%+ higher IPC in the near future, and you can’t expect clocks to fix this when they already is near the roof here. SB is capable of the same frequencies and is so on much lower power usage and with much higher IPC, despite half the core count! Intel has more than twice the performance per mm² at much lower power usage, you can’t simply fix that with a few revisions. Phenom II has higher performance per watt, twice(!) the performance per mm² (taking processes in to account) and higher IPC and is capable of almost the same- if not the same or higher - frequencies on the same process.

The architecture is seriously broken, much more so than P4 ever was! And at the same time AMD has put so much pride and marketing in this that they can’t go back, they’re stuck with this mess for years!


Ask yourself, when do you think this mess i solved? You can sell products with subpar performance given they are cheaper and/or cooler. That's what AMD did in the past when they was behind, with the Phenoms, Athlon X2, K6 and K5. Bulldozer can't even do that!
Me being an AMD "fanboy" since K6 is now planning for Ivy Bridge to replace my Phenom II, the main reason I didn't like intel in the past was that they made inefficient processors with high frequency and power usage without the performance how do you thin BD or PD will stick up to IB in those areas?

[TESKATLIPOKA]

-Boris-

Phenom II has higher performance per watt, twice(!) the performance per mm² (taking processes in to account) and higher IPC and is capable of almost the same- if not the same or higher - frequencies on the same process.

can you show me some proof, because some parts are so wrong, mainly 2x performance per mm2, I really want to know how you got that and the same or higher frequencies is just a dream, its enough if you look at Llano with disabled IGP, that cpu can't break the 4Ghz barrier while BD is attacking 5Ghz and is made on the same 32nm process.

p.S. I agree on higher IPC and perf/w is worse only on 4module but still on par with Thuban, the 3module and 2 module are better than anything else from AMD. Example Llano 2.9Ghz has the same power consumption as FX4100 but FX is a few % better.
Even FX6100 has a higher consumption by just 4% but performs better by 15% the rest of AMD cpu has higher power draw.

[Oliverda]

Phenom II has higher performance per watt, twice(!)

Bulldozer: 315 mm2

Thuban: 346 mm2


Just to add that GF's 45 nm process is capable for higher performance than the current 32 nm one.

[Smartidiot89]

Because we are not too fond of turd polishing! It’s a botched architecture, you can’t just give it 50%+ higher IPC in the near future, and you can’t expect clocks to fix this when they already is near the roof here. SB is capable of the same frequencies and is so on much lower power usage and with much higher IPC, despite half the core count! Intel has more than twice the performance per mm² at much lower power usage, you can’t simply fix that with a few revisions. Phenom II has higher performance per watt, twice(!) the performance per mm² (taking processes in to account) and higher IPC and is capable of almost the same- if not the same or higher - frequencies on the same process.

I am not going to say Bulldozer is great cause it isn't. Facts are it was designed with power efficiency in mind and something has gone terribly wrong with the architecture, and we won't know for sure if these problems can be resolved until we see Piledriver to be honest.

And clocks aren't "near the roof" - far from it. AMD have for the past two quarterly results said explicitly they aren't happy with 32nm performance at GlobalFoundries and have overall been very open about it. Bulldozer was designed with high frequencies which shows as it retails at 3,6 GHz base. Llano was aimed at >3,0 GHz and reached only 2,9 GHz at launch and lets not speak of the mobile models. There are no problems releasing a CPU with clocks above 4,0 GHz and good power efficiency as long as it was designed for it (IPC usually gets cut then) which was a design compromise AMD did.

The manufacturing process isn't good at all today, and AMD decided to launch a brand new architecture on an unproven manufacturing process (never done by AMD nor Intel ever before?). I will wait for Piledriver before I pass any judgment on the architecture as a whole. I don't expect it to rock anyones world, but it will most likely be better than K10 and Phenom II.

[Piledriver]

Overall faster than Thuban? Someone forgot to give me that memo?

Don't know about the memo, but how about every review available?

[FlanK3r]

I must say it again. No, FX is definetly not great product now, but simultaneously not bad product. SOmetimes looks good in practice, sometimes whorse. I think, new architecture can chance from forum users, we will smarter in enxt years, if Bullodzer was wrong step or not (Piledriver, Steamroller)

[BeepBeep2]

it reminds me that you are still a child can you act more sensible, please?

saying all time how AMD should to do, like you know better than all engineers of amd alltogether. Ironic?

It doesnt make any sense, get of the past and act in current reality. Top priority for amd is get a competitive Opteron. They fail or not, it is an opinion of everybody and it just takes time to optimize softwares and it goes quickier there in server environment. Complaining about desktop performance is understandable.

Let me show you some charts.

Desktop performance from four different review sites (what I've been complaining about)
IPC/Single Thread:






Gaming (CPU bottleneck):







Power consumption:
(what the is that?)



Even Phenom (I) took a step forward in IPC over 10%, added two cores over the architecture it replaced (at the expense of clockspeed/TDP), overclocked as well as the architecture it replaced and you all are considering it much more a fail than Bulldozer.

You know, it seems I am skewing results a bit, hand picking charts and whatnot.

Here's a good one. FX does relatively well in x264.

Now here is the accompanying power consumption chart.

Did I mention that FX's temperatures are through the roof? A lot of people are hitting 80c+ on air (unknowingly because the temperature diode is broken reading 15c+ low) overclocked...its rediculous.

Seriously, I have no idea why people are still crying about Bulldozer. Sure it's not the fastest chip,but at least it is overall faster than Thuban(sometimes a lot ,sometimes it's even slower). The good thing is that even in this worst case scenario when both design and process have some issues,it's quite a decent product. It just needs a price correction to be a bit more competitive from price/perf. POV.
Piledriver will improve on Bulldozer,Steamroller on Piledriver etc. The curve goes to the right direction so I don't know why some people are so pessimistic... If anyone thinks that AMD could have rode K10 on sub 32nm with 4+GHz clocks and 8+ cores with no changes to the uarchitecture itself is delusional.

Charts are above.
I never said AMD could have rode K10 in on 32nm with 4 Ghz clocks and 8 cores, that would beat even Sandy Bridge up by a large margin.
6 K10 cores, BD's memory controller on 32nm with 3.6 stock frequency and 4 Ghz Turbo, not doable?

Look over your posts and then tell me your not pulling stuff out of thin air. If not, then show us the "facts" you so boldly claimed?

Charts are above. They may be handpicked (you asked me to find them...)...

Because we are not too fond of turd polishing! It’s a botched architecture, you can’t just give it 50%+ higher IPC in the near future, and you can’t expect clocks to fix this when they already is near the roof here. SB is capable of the same frequencies and is so on much lower power usage and with much higher IPC, despite half the core count! Intel has more than twice the performance per mm² at much lower power usage, you can’t simply fix that with a few revisions. Phenom II has higher performance per watt, twice(!) the performance per mm² (taking processes in to account) and higher IPC and is capable of almost the same- if not the same or higher - frequencies on the same process.

^^
This is pulling out of thin air. Well, some of it at least. His comments in the middle about SB capable of same frequencies is true, much lower power usage is true, much higher IPC is true, despite half the core count is true. His comment about twice the performance mm^2 (sandy bridge only) needs to be checked though, but considering that Sandy Bridge has integrated video, we need to look at that closer and remove that part of the die from the equation.

As far as his comments about Phenom II, it does get higher performance per watt, more performance per mm^2 (pretend AMD released Phenom II X6 on 32nm, it would beat BD by far in a performance per mm^2 standpoint) but not by a factor of 2x. More like 1.4x. It also has 15-20% higher IPC...and who knows if they couldn't have gotten the same or better clocks on 32nm? The chip would run cooler, use less wattage and probably gain a few hundred Mhz headroom unless you really think that GloFo's 32nm is WORSE than their 45nm was.

[Zeus]

Don't know about the memo, but how about every review available?

Look at the post above. Don't see it being any faster than Thuban, do you?

[BeepBeep2]

Look at the post above. Don't see it being any faster than Thuban, do you?

*pulls out WinRAR charts, CS5 chart and some random AES encryption chart comparing only Thuban and Zambezi as well as x264 chart, comparing only Thuban and Zambezi and 2500K, excluding the power consumption chart*
(Because we all use these daily, all the time...I know I have to make archives everyday...)

Sure it is!

Too bad my games would all be slower :'(
(*pulls out BF3 chart*)
Nevermind!

...and my web browser...and my other single threaded programs, look at GIMP in the bit-tech review, just look at the iTunes and Lame chart above, I don't care if you can do that at 5 Ghz with BD it will still turn out slower than Thuban at 4.2. Look at Cinebench scores in single thread (oh wait, that is a bad representation of BD's performance...)

I'm waiting for wez and Tomasis to chime back in again about how I'm so full of it, how immature I am, how insensible I am, they will go handpick benchmarks just like I did about BD's performance gains, however they are usually by a small margin (~10% like I said in my 'full of it' post) and do indeed take advantage of all 8 of BD's clusters because IPC is so bad.

[FlanK3r]

u are rendering in single core mode? ,-)

[Smartidiot89]

As far as his comments about Phenom II, it does get higher performance per watt, more performance per mm^2 (pretend AMD released Phenom II X6 on 32nm, it would beat BD by far in a performance per mm^2 standpoint) but not by a factor of 2x. More like 1.4x. It also has 15-20% higher IPC...and who knows if they couldn't have gotten the same or better clocks on 32nm? The chip would run cooler, use less wattage and probably gain a few hundred Mhz headroom unless you really think that GloFo's 32nm is WORSE than their 45nm was.

I will reiterate what I said above: AMD have explicitly said they are not happy with GlobalFoundries 32nm performance and are working with them on fixing it. They've said this on quarterly reports two quarters in a row. Llano was supposed to launch at frequencies above 3,0 GHz and it did so at 2,9 GHz - the mobile models aren't even funny to describe. Llano isn't a train wreck like Bulldozer, but it confirms there is something wrong with the process - just like AMD's own words from people like Thomas Seifert.

I think AMD's engineers knew exactly what they were doing when they designed Bulldozer, it's foolish to even question this. However the end result is bad, probably for many reasons but the architecture is in no manner "bad". The manufacturing process on the other hand is a dog at the moment. AMD decided to build an entirely new architecture on a brand new and unproven process, a very dangerous combination the semiconductor industry usually never attempts (also a strong reason for Intel's Tick-Tock scheme).

[Tomasis]

Beep, OMG. cannot you read and think logically interpreting my post?

Im not looking for what should be done or not.

Honestly, do you think AMD cares about high end CPU for desktop users? That gives relatively small earnings per cpu in that nische market. What matters most are Brazo, Lllano and Opteron from economic viewpoint for AMD. Large volumes for large base of users that dont OC.

I said BD is designed for Server environment at first hand. Check Cray machines that are equipped with Nvidia cards. Calculations are needed of combined BD and Nvidia hence double fpu of Bd or whatever it is. Times are changing and GPU take more use of calculations. So I assumed that the new uarch is designed for a few years forward with eventual software upgrades and workloads divided to different things.

So the first gen is miss or luck with relatively small investments from AMD side and I dont need mention how much money one need put down for succesfull lower nm manufacturing like GlobalFoundries. Compare numbers with Intel investments in $$$$ It is very large sum. Ticktock sounds better and safer but such is reality when you have take a chance and try out it on new ground.

Didnt I mention that complaining about desktop perf is understandable? It means no much comments about it from my side, it neither is great or fail rather a last resort for desktop after Opteron (likewise with Phenom). I'd get more use of new BD since I have 955, for my work I use Photoshop CS5, 3d rendering, after effects. It makes sense for me when I need 100% utilizition of all cores, not single thread perf. Myself I see BD as 4 core cpu with 8 threads so it is a leap forward when I upgrade it from 955. It'd be nicer if I save a few watts energy consumption but I can live with it. If you look at Phoronix Linux benchmarks, it confirms my suspicions how and what purpose the new architecture is designed for.

If you are not happy, why not buy 2600k, it is not too late sell your Crosshair V

I prefer constuctive attitude towards all things in life, it is waste time to write things when it is all in past anyway.

[muzz]

Not sure how anyone can blame GF for the architecture failure.
Granted, they DO take up too much power, and it DOES heat up too much when OC'd....

What else can be THEIR fault?
How fast was it supposed to be, 6G? How was the performance of that monster OC that was on here, did it even compete with a 2600 or 990 at 4.5 g?
At stock frequencies the draw and heat are fine, and this chip gets WHALED on, which tells me and anyone else paying attention that's it's IPC is weak.
We still live in a ST world, regardless of what the geniuses over at AMD say.

[muzz]

Not sure how anyone can blame GF for the architecture failure.
Granted, they DO take up too much power, and it DOES heat up too much when OC'd....

What else can be THEIR fault?
How fast was it supposed to be, 6G? How was the performance of that monster OC that was on here, did it even compete with a 2600 or 990 at 4.5 g?
At stock frequencies the draw and heat are fine, and this chip gets WHALED on, which tells me and anyone else paying attention that's it's IPC is weak.
We still live in a ST world, regardless of what the geniuses over at AMD say.

[BeepBeep2]

Beep, OMG. cannot you read and think logically interpreting my post?

Im not looking for what should be done or not.

Honestly, do you think AMD cares about high end CPU for desktop users? That gives relatively small earnings per cpu in that nische market. What matters most are Brazo, Lllano and Opteron from economic viewpoint for AMD. Large volumes for large base of users that dont OC.

I said BD is designed for Server environment at first hand. Check Cray machines that are equipped with Nvidia cards. Calculations are needed of combined BD and Nvidia hence double fpu of Bd or whatever it is. Times are changing and GPU take more use of calculations. So I assumed that the new uarch is designed for a few years forward with eventual software upgrades and workloads divided to different things.

So the first gen is miss or luck with relatively small investments from AMD side and I dont need mention how much money one need put down for succesfull lower nm manufacturing like GlobalFoundries. Compare numbers with Intel investments in $$$$ It is very large sum. Ticktock sounds better and safer but such is reality when you have take a chance and try out it on new ground.

Didnt I mention that complaining about desktop perf is understandable? It means no much comments about it from my side, it neither is great or fail rather a last resort for desktop after Opteron (likewise with Phenom). I'd get more use of new BD since I have 955, for my work I use Photoshop CS5, 3d rendering, after effects. It makes sense for me when I need 100% utilizition of all cores, not single thread perf. Myself I see BD as 4 core cpu with 8 threads so it is a leap forward when I upgrade it from 955. It'd be nicer if I save a few watts energy consumption but I can live with it. If you look at Phoronix Linux benchmarks, it confirms my suspicions how and what purpose the new architecture is designed for.

If you are not happy, why not buy 2600k, it is not too late sell your Crosshair V

I prefer constuctive attitude towards all things in life, it is waste time to write things when it is all in past anyway.

I wasn't complaining about Opterons though, I simply mentioned them once in reference to how badly they did in the anandtech review. They lost to intel's equivilents while consuming more power than the last generation and lost to the last gen 2/3 times.

If their goal was many slow cores in a server environment why not just glue more on the old uarch?

Intel continues to make single cores faster so it can use less of them or clock them down when they add more.

EDIT:
I'd like to note that you keep downplaying my intelligence as if my age is what causes my percieved stupidity. You've so far told me that I can't think logically, can't interpret simple one sentence posts, you've told me to "act more sensibly" even. All because "you remember that I am a child." Your own descrimination puts me a step down before I even say a word.

You thanked a guy for trolling me. Get real.

[demonkevy666]

why do you keep talking about IPC
it's IPM LOL module shared L1 instruction cache people.

[muzz]

Module Smodule.
I just prefer to call it slow doing what it is supposed to do>

Compute.

[muzz]

Module Smodule.
I just prefer to call it slow doing what it is supposed to do>

Compute.

[BeepBeep2]

http://arstechnica.com/business/news...tastrophe.ars/

We are all off topic, this thread was designed to speculate on Piledriver which is hopefully 40% faster if they can find 30% in the process refinement and 10% in the architecture by then.

I guess 45nm improved about 30% since they launched Phenom II X4 940 so hopefully 32nm does that too. It needs to do it quickly.

[muzz]

I thought they refined the core AND the process got better?
If they can do the same with BD it MAY turn out OK, I have my doubts.....but I'm hopeful.

[tbone8ty]

yeah Amd needs to do something stat.

release b3 stepping asap

or heck make yur top of the line bulldozer available for once lol

man good thing u can still buy their phenom II line up otherwise they'd be SOL