AMD FX-8150 Bulldozer finally tested

[Andrew LB]

If I worked at some place and say in public a piece of data that is not protected under NDA (While I know all the other details I can NOT say to be able to reach and sustain such a conclusion), you are indeed going to believe me much, much more that some random guy with an Engineering Sample leaking results. I don't think he could have stated something without knowing if it was true or false as he knew all the details we didn't, this is why I don't get why he could have insisting soo firmly in the "IPC increases" thing if it wasn't the case. Maybe he was comparing a Bulldozer module against a single K10 Core, that is the only way you can make sense of it.
Also, he was quite accurate with the statement that Bulldozer wasn't compatible with the standard AM3 platform and at that moment I think most believed that it could been a drop-in replacement.

Now you're just making excuses for AMD.

I think it's more than likely that Jeff was simply fed inaccurate information from people higher up in the company... which leads me to believe one of two things. Either management is completely clueless about the architecture of Bulldozer and internal performance estimates, and due to incompetence would just tell Jeff what he wanted to hear to hurry his departure from said office. OR.... someone in management was FULLY aware of how Bulldozer would perform and perhaps didn't really like Jeff very much, thus giving him incorrect date AND putting office politics before the image of the company.

Neither are that far fetched. I personally think the latter is most likely considering I once was in that exact same situation a few years back.

[blindbox]

He asked his engineers directly.

It's more likely the reply he got were solely based on synthetics and also the newest instruction sets, or on server workloads.

It's John btw.

[Dresdenboy]

He asked his engineers directly.

It's more likely the reply he got were solely based on synthetics and also the newest instruction sets, or on server workloads.

I think that engineers building a new microarchitecture won't measure performance or IPC by executing 10h, NHLM, Atom or 386 traces. They would use SPEC or other workloads as it is common for scientific research and assume it to be compiled for the target hardware. This could well be the base for when they are talking about IPC.

[Opteron146]

I think that engineers building a new microarchitecture won't measure performance or IPC by executing 10h, NHLM, Atom or 386 traces. They would use SPEC or other workloads as it is common for scientific research and assume it to be compiled for the target hardware. This could well be the base for when they are talking about IPC.

That's probably the main problem. The engineers talked about Server & HPC IPC, i.e. running typical server programs. People often forget, that IPC is software/program depended. Seems that AMD knew that from the start:
1_amd_bulldozerhkgt.jpg
The performance plus is the smallest for the desktop segment. That there is now no plus at all is not a hugh error margin.
Maybe JF and/or his engineers forgot the performance differences in the different fields/applications. As he always used to say, he is a server guy - not desktop person.

Another side note: The above graphics is before the K10 launch, thus the "plus" is probably referring to the K8, only. However, I guess that the slide was meant for the never released Bulldozer in 45nm, which was once upon the time on the road map, but got canceled later.

[Dresdenboy]

And I don't see a lot of consumer grade applications getting optimised for BD, since it's made by AMD and its performance is poor.

I couldn't resist this chicken-egg problem So you mean software won't get optimized or recompiled to run e.g. 20% faster since it currently runs 10% below expectations.

[Manicdan]

I couldn't resist this chicken-egg problem So you mean software won't get optimized or recompiled to run e.g. 20% faster since it currently runs 10% below expectations.

i think he means because it affects a smaller portion of the market too.

[Manicdan]

I couldn't resist this chicken-egg problem So you mean software won't get optimized or recompiled to run e.g. 20% faster since it currently runs 10% below expectations.

i think he means because it affects a smaller portion of the market too.

[zalbard]

I couldn't resist this chicken-egg problem So you mean software won't get optimized or recompiled to run e.g. 20% faster since it currently runs 10% below expectations.

BD performs poorly => less people buy it => BD is a part of smaller client base => BD is not statistically relevant to put effort into optimising for it.

[kam03]

AMD knew BD was a flop, this must be the reason why CEO got fired, resigned or what ever the case was?

[Dresdenboy]

Or this explains it:

http://www.xbitlabs.com/news/cpu/dis...er_Fiasco.html

C. Maier left the company many years ago. Tools for automated design have extensively be used for Bobcat (it's a synthesizable design) and for 10h, 11h, BD, the GPUs and so on. But all that in different amounts. The BD ISSCC papers mention explicit optimization of critical paths if there were some. But this is not the same as a fully synthesized design (like Bobcat), where AMD engineers stated 20% lower performance and 20% more area. Xbit labs gets everything wrong somehow.

Further background info:
http://forums.anandtech.com/showpost...3&postcount=82
http://forums.anandtech.com/showpost...2&postcount=91

[Dresdenboy]

BD performs poorly => less people buy it => BD is a part of smaller client base => BD is not statistically relevant to put effort into optimising for it.

Less people buy it at those prices.

Just got a slide set into my hands. So for 2012 GF promised to increase 32/28nm output by about 2.5x vs. 2011. With that, the smaller 4c BDs and especially Trinity as a mass product (replacing Llano) and Piledriver-based desktop chips, AMD might still gain some significant share. But my initial answer wasn't meant to be that serious (thus the smileys).

[zalbard]

Less people buy it at those prices.

Just got a slide set into my hands. So for 2012 GF promised to increase 32/28nm output by about 2.5x vs. 2011. With that, the smaller 4c BDs and especially Trinity as a mass product (replacing Llano) and Piledriver-based desktop chips, AMD might still gain some significant share. But my initial answer wasn't meant to be that serious (thus the smileys).

I really hope they do work things out and that PD is a success. I was looking forward to BD until I saw the benches...

[Sparky]

I really hope they do work things out and that PD is a success. I was looking forward to BD until I saw the benches...

I agree! My 955 is getting long in the tooth... Though it still worked out as I don't have the cash right now to upgrade anyway, even if BD was an uber amazing chip. Maybe PD will be great and I'll have money then. Maybe.

[Cecil]

Why is everyone wanting PileDriver? What Ive found is that PileDriver is the replacement for Llano. Bulldozer Enhanced is the update version of BD that will work on 990 series boards.

[EniGmA1987]

Why is everyone wanting PileDriver? What Ive found is that PileDriver is the replacement for Llano. Bulldozer Enhanced is the update version of BD that will work on 990 series boards.

Um, no. Piledriver includes 2nd gen Bulldozer. Steamroller is 3rd gen BD

[RedBull78]

Hey enigma can I have your old system?

[Opteron146]

Why is everyone wanting PileDriver? What Ive found is that PileDriver is the replacement for Llano. Bulldozer Enhanced is the update version of BD that will work on 990 series boards.

2 names, same thing. I prefer to refer to it as BDv2, it's the shortest version ;-)

[Cecil]

Um, no. Piledriver includes 2nd gen Bulldozer. Steamroller is 3rd gen BD

That chart is only showing two things. First, performance per watt, not overall performance, and second, when new CPUs/APUs are launching. Says nothing about platforms.
For PileDriver it even says "APU Trinity".

2nd Generation

AMD Financial Analyst Day 2010,[27] revealed the 2nd generation is to be scheduled for 2012. AMD currently refers to this as Enhanced Bulldozer.

This new generation of Bulldozer core is codenamed Piledriver and will be incorporated into specific desktop and notebook markets:
Desktop Performance market (Volan platform[28]): Zambezi will be replaced by Vishera (up to 8 cores). Vishera will feature Turbo Core 3.0 while using the existing Socket AM3+ format and 9xx series chipset of the 1st generation FX-series Zambezi processor. AMD has projected this 2nd generation FX-series processor will be 10% better under digital media workloads.[29]
Desktop Budget and Mainstream market (Virgo platform[30]): The Stars-based Llano Fusion APU line will be replaced by Trinity, Weatherford, and Richland Fusion APUs (2 to 4 cores) that are aimed at various price points in the desktop market. This platform will use Socket FM2 format.[31] Notebook Mainstream and Performance market (Comal platform[32]): Will be the same as mentioned in Desktop Budget/Mainstream market.

At AMD Fusion Developer Summit (AFDS) 2011, AMD has claimed the computational capacity of the notebook variant of Trinity will be 50% faster than Llano.[33][34][35]

http://en.wikipedia.org/wiki/Bulldozer_(processor)

[duron]

costumer satisfaction

http://www.tigerdirect.com/applicati...y=#ReviewStart

[BeepBeep2]

That chart is only showing two things. First, performance per watt, not overall performance, and second, when new CPUs/APUs are launching. Says nothing about platforms.
For PileDriver it even says "APU Trinity".



http://en.wikipedia.org/wiki/Bulldozer_(processor)

Yes, the Trinity APU includes Piledriver cores.
Zambezi is to Bulldozer as Vishera is to Piledriver.

[EniGmA1987]

Hey enigma can I have your old system?

Sorry just sold it not long ago.

That chart is only showing two things. First, performance per watt, not overall performance, and second, when new CPUs/APUs are launching. Says nothing about platforms.
For PileDriver it even says "APU Trinity".
http://en.wikipedia.org/wiki/Bulldozer_(processor)

I had said Piledriver INCLUDES the next bulldozer. It also includes the next Llano. Like the slide says. And thanks for posting the wiki article, it also confirms what I said But also gives the actual core name which I was not yet aware of.

[Andrew LB]

He asked his engineers directly.

Hmm. Bulldozer taped out almost 15 months ago. That being the case, they've had working silicon for over a year now, so I think the idea that this was all based on synthetic software simulations of processor functionality do not hold up to the evidence. Like I said earlier in one of these threads, everything that John (Sorry John!!) talked about in regards to Bulldozer in the past year and a half was without question approved for public consumption. Such info demands close verification and scrutiny, two things that those in management didn't seem to bring to the table. IMO some of the suits in upper management are guilty of complete incompetence. I have a feeling in a few weeks we'll be seeing a few people 'moving on to better things' or 'spending more time with the family'.

And now that he knows what it's like to get burned when you rely on blind faith, I'm sure John will not become complacent. And maybe he'll get a nicer office after the coming evictions.

It's John btw.

Again, my apologies. Spent far too many hours hanging drywall in my garage today. Time for sleep.

[BoredByLife]

http://semiaccurate.com/2011/10/17/w...er-underwhelm/

http://semiaccurate.com/2011/10/17/b...ingle-problem/

Interesting read

[PatRaceTin]

still no retail in Thailand

[haylui]

still no retail in Thailand

FX has arrived in Singapore