AMD Bulldozer Thread

[rouge]

I can't beleive what a steaming pile of crap AMD had the balls to release. How in the hell can it be slower clock for clock in single and multi threaded programs than Thuban even with 2 more cores and be more power hungry? They should just hang it up and fold since there's no way anybody in the right mind is going buy these inneficient slugs. No wonder they made AM3+ boards available so early. There going to have to slash prices right away if they want to sell anything. I still just can't believe they went ahead and luanched these turds it totally blows my mind how poorly they perform. Everyone at AMD should be ashamed of themselves Intel must be luaghing there asses their at your incompetence and ignorance.

[HelixPC]

Damn, this is almost a bigger flop than the original Phenom and the 2900XT combined

[JohnZS]

Well, Ket, me and quite a few other members of the XtremeSystems forums kept warning AMD fans not to whip themselves up into a FRENZY and to MANAGE THEIR EXPECTATIONS. This is EXACTLY what I expected.
For some reason if you turn back the clock ~9years the role is reversed. Intel had the Netburst and rather hot Prescott, AMD had the efficient and rather fast Athlon A64.
Core 2 Duo was the turning of AMD and Intel's role in the market place and amongst enthusiasts.

I see no reason to purchase a Faildozer... sorry Bulldozer. i7 2600k is looking more and more fruitful, however I am still waiting for Sandy Bridge-E before I upgrade from my SKT-775 platform.

I wish AMD would not hype so much before launch and that AMD Fans would not whip themselves up into a Frenzy. WHY? Because if appropriately priced this could be a very decent multi threaded crunching platform (especially on a multi-socket board).

In my opinion Fusion is the future, especially for low profile HTPC's For gaming and any serious CPU power Intel completely dominate.

John

[sutyi]

Bulldozer has one big issue. GloFo's 32nm node.

Not high enough base and turbo clocks due to epic leakage.

[tajoh111]

The scariest thing about this architecture is that it seems like it has no long term viability. High clocks have been not the way to go for performance ever since laptops started outselling desktops.

If your putting bulldozer in a laptop, your both going to get bad performance and bad battery life. Even if they fix the manufacturing process, they will needs to boost speeds well above 3.5 ghz and this will be a battery killer. The key to getting long battery life, speed and thus mobility is high performance with low clocks. Months ago, before all this performance numbers were released, Bulldozer I thought had too high of clocks to get competitive performance.

This thing is in many ways worse than fermi. Fermi atleast took the crown and was much faster than its predecessor. It also appears to have an easier fix, as components that were already in the architecture were disabled on earlier models and reenabled for it forthcoming models when the bugs were kinked out.

Bulldozer is fully enabled and adding more clocks is a band aid solution because they are going to run into a wall where adding more clocks adds to much heat and power consumption.

AMD used to be about more performance per hertz. I miss those times. Remember when a 1800+ thunderbird performed like a 2 ghz. intel even though it was like 1.5 ghz or something. What happened to those days. AMD has lost focus so much. I hope AMD has a new design in the works, but it looks like from their 10-15 percent performance increase over years, we are stuck with BD.

The biggest reason for disappointment for me, is BD was supposed to be and had to be the foundation which AMD builds the rest of their company. With BD being so bad from the get go, I have a feeling AMD may concede the high end market altogether(they must hate selling such a big chip with so much cache for so little) and basically drain until it is dry, their APU product, which will dry up sooner than later(next year) when it doesn't bring enough CPU performance to the table. I could see AMD not being around in 4 years.

[M.Beier]

My updated signature describes the problem perfectly...
It is all the software that is to blame, it is actually a product that will beat Intel's flagship for 2015, really.

[JohnZS]

I could see AMD not being around in 4 years.

No, Fusion is doing extremely well and if AMD continue to make strides in this sector then I can see it becoming even more fruitful for AMD so AMD are not dead..... yet.. or even in 4 years time.
Also lest not forget that they are not doing too bad with their Graphics Cards either.

John

Heheh thnks M.Beier for a bit of light relief in this thread

[M.Beier]

No, Fusion is doing extremely well and if AMD continue to make strides in this sector then I can see it becoming even more fruitful for AMD so AMD are not dead..... yet.. or even in 4 years time.
Also lest not forget that they are not doing too bad with their Graphics Cards either.

John

Heheh thnks M.Beier for a bit of light relief in this thread

Yea, I mean, why try to dig the hole deeper.... It is a failure, AMD will rise and shine again, you win some and you loose some - this is not the deathstroke for AMD in any way whats all ever, and particular in SFF they are kicking @ss on Intel.... But pretending this is really what we expected, and this is not fail, I mean, come on...
As many others before me have expressed, it does hurt that they, AMD, choose to abuse the FX legacy.

[Leeghoofd]

So you think the power consumption numbers are higher because people used the stock coolers and more voltage than they could have?

At my knowing there were not any coolers included in the early press kits. New kits maybe have a what looks like a rebranded Antec all in one cooling solution.

[marten_larsson]

Loved the Hitler video, laughed hard when I watched it late last night, just as good as the Fermi video. My girlfriend didn't understand what was so funny, I didn't care to even try to explain.

[n!Cola]

Somehow the only one with FX-6100 review. Croatian site but it is easy to see one more fail from amd, only much lower price will help...
http://www.pcekspert.com/articles/905.html

[onethreehill]

Legion Hardware and Guru3D reviewed FX-6100

AMD FX 8150 - 8120 - 6100 and 4100 performance review
http://www.guru3d.com/article/amd-fx...rmance-review/

AMD FX-8150 at 4.7GHz. Does it stand tall?
http://hexus.net/tech/reviews/cpu/32...es-stand-tall/

[Chrono Detector]

Wow, the FX-6100 sucks so bad even the 1055T is a better option.

[ajaidev]

Yesterday after talking a lot about BD with a few colleagues an simple conclusion was reached. But when i came to post it the forum was in a 1Hr trance that did not seem to end :P

What was concluded was that since this is a totally new arc there were bound to be issues. BD 1st gen was never expected to be a Intel processor beater by Intel staff. You see it all boils down to how much experience have you got dealing with that arc now Intel has been working on the same base arc for quite some time "Derivative of c2d". All these Nehalem and SNB, etc carry a lot less risk than a full fledged different arc like new FX to the old Phenom 2.

You all might recall the problem that P4 had faced with efficiency and coupled with HT it was even worst. Well engg. concluded that longer pipes were off the table till a small enough process is adopted that can take advantage of it and we will see longer pipelines in future Intel processors for sure. Now before people start saying Haswell is completely new well there has been some changes to the original plan.

What is wrong with BD that everyone agrees with is not the lack of performance but the electricity used on full load. A A8 cpu seems a better alternative to a two module BD with 4 cores. If sufficient improvement is not done with the pile-driver we may have similar results as the A8's in light process application.

The other small bit is that i heard AMD was about to use the phrase "4+4 core" instead of the "8 core" moniker which in my mind would have been much better. Also the water cooler was suppose to be bundled with a faster processor but due to problems with yield sufficient speed bin could not be reached but we will surely see a cheap FX 8200 "Or something" with a bundled water cooler.

[Olivon]

wow this thingie is slower and hotter than ancient yorkfield in gaming..

[img]http://www.abload.de/img/img0033850nuhi.gif[img]
Indice 100 = Deneb

The worst part of HFR review :



http://www.hardware.fr/articles/842-...ergetique.html

Even a Q6600 is more efficient than others AMD processors reviewed

[kaktus1907]

The worst part of HFR review :

http://tof.canardpc.com/view/55d9fbb...0dcad33ec8.jpg

http://www.hardware.fr/articles/842-...ergetique.html

Even a Q6600 is more efficient than others AMD processors reviewed

i remember this from somewhere else.. slower, hotter, expensive and late FAIL⁴

[schlafmuetze]

I dont have the Time to read through all reviews but, is one of those reviews with a Gigabyte 990FX Board? I just checked if there was a Bios Update for my MB and for real yo - there is.
Only an Updated CPU Agesa Code tho - which could mean improved BD Performance?(Thats why im posting here)
Quick google brought nothing up

Link to teh Downloadpage: http://www.gigabyte.de/products/prod...?pid=3891#bios

[kl0012]

I wasn't really surprised by the performance of Bulldozer. But AMD still managed to surprise me.
BD has 2x more transistors and 30% larger die then SB (and this is without moving SB IGP & SA out of the formula)? Common, what are all these transistors used for? Maybe BD has some secret functions we have no clue about?
Definitely it's not AVX. Because BD AVX implementation - and this is another surprise from AMD - seems even slower then SSE.
http://www.lostcircuits.com/mambo//i...&limitstart=13

[Hornet331]

AMD FX-8150 at 4.7GHz. Does it stand tall?
http://hexus.net/tech/reviews/cpu/32...es-stand-tall/

Well at 4,7ghz performance doesn't look that ba.... holly balls that power consumption.

I don't want to know the power consumption in those 8ghz ln2/lhe bench sessions...

I wasn't really surprised by the performance of Bulldozer. But AMD still managed to surprise me.
BD has 2x more transistors and 30% larger die then SB (and this is without moving SB IGP & SA out of the formula)? Common, what are all these transistors used for? Maybe BD has some secret functions we have no clue about?
Definitely it's not AVX. Because BD AVX implementation - and this is another surprise from AMD - seems even slower then SSE.
http://www.lostcircuits.com/mambo//i...&limitstart=13

There happens a lot of discussin about that, many say on of the reasons is that happend because of the heavy use of automated desgin tools and next to no hand optimization happend. If you look at what we got and when you look at what Cliff Maier said back in 2010:

I don't know. It happened before I left, and there was very little cross-engineering going on. What did happen is that management decided there SHOULD BE such cross-engineering ,which meant we had to stop hand-crafting our CPU designs and switch to an SoC design style. This results in giving up a lot of performance, chip area, and efficiency. The reason DEC Alphas were always much faster than anything else is they designed each transistor by hand. Intel and AMD had always done so at least for the critical parts of the chip. That changed before I left - they started to rely on synthesis tools, automatic place and route tools, etc. I had been in charge of our design flow in the years before I left, and I had tested these tools by asking the companies who sold them to design blocks (adders, multipliers, etc.) using their tools. I let them take as long as they wanted. They always came back to me with designs that were 20% bigger, and 20% slower than our hand-crafted designs, and which suffered from electromigration and other problems.

And when you look back, how furious AMD fans where at this, calling him a disgruntled ex-employee who only wants to put amd into a bad light...

[Kuroimaho]

Well at 4,7ghz performance doesn't look that ba.... holly balls that power consumption.

I don't want to know the power consumption in those 8ghz ln2/lhe bench sessions...

Compare it to the 2600K...

No doubt Bulldozer at 4.7 24/7 will be popular.

[mav2000]

Im already seeing a whole lot of motherboards being RMA'ed due to burnt up VRM area.....

[ajaidev]

Bloody hell Newegg has the FX8150 for $280 wtf?

http://www.newegg.com/Product/Produc...82E16819103960

While i am not very happy with FX8150 the FX8120 if priced around $200-$210 will be a good deal because Intel 2400 costs around $190 on newegg.

[Hornet331]

Well.. extremly limited supply and an automated pricing system...

And damn.. double post everywhere again... I thought the downtime was there to fix that.

[ice_chill]

Well when the first generation Pentium 4 came out, it was very expensive and clock for clock was slower against Pentium III, but after a few revisions it stood up to the top model Athlon XP 3200+. Then later Nehalem was released, inheriting Hyperthreading which was developed thanks to Pentium 4.

There is no reason why tweaking cannot get the Bulldozer on track, remember it has excellent multithreading performance, just the IPC that needs to be tweaked.

Also AMD engineers are speaking out that they are underfunded by the new management which is why they don't have the resources to create a good chip.

[Hornet331]

Later moddels of P4 fought against A64... and it got clobbert, regradless how high you oced them... hell people even bought Pentium-Ms oced it and beat the crap out of P4s that where oced much higher...