AMD FX: Energy Efficiency Compared To Eight Other CPUs

[buzznut]

The FX 8150 has been tested over and over, we all know what it does and what its strengths and weaknesses are.
Bulldozer is more than just the 8150, so I'd like to see more tests run on some of the other CPU's. For comparison if nothing else. Especially interesting is the FX4100.

 

[dragonsqrrl]

[citation][nom]AMD fanboy[/nom]but that is not performance per watt[/citation]
umm... yes it is. Performance per watt is essentially a measure of power efficiency, and that's the exact comparison the bottom chart makes.
[citation][nom]AMD fanboy[/nom]so many reviews and so different tests for multitaskingheres 1 rating fx 8150 with 3/5 stars http://www.expertreviews.co.uk/pro [...] -fx-8150/2[/citation]
Here you go...

http://semiaccurate.com/

It's hilarious that you're trying to reinforce your argument with a review that gave the FX-8150 a 3/5. Is that the best cherry picking has to offer?

And again, you might be more comfortable here:

http://semiaccurate.com/

 

[dragonsqrrl]

[citation][nom]buzznut[/nom]The FX 8150 has been tested over and over, we all know what it does and what its strengths and weaknesses are.Bulldozer is more than just the 8150, so I'd like to see more tests run on some of the other CPU's. For comparison if nothing else. Especially interesting is the FX4100.[/citation]
Guru3D did a review of the entire FX lineup:
http://www.guru3d.com/article/amd-fx-8150--8120-6100-and-4100-performance-review/

At its current price of $130 it honestly isn't worth it:
http://www.newegg.com/Product/Product.aspx?Item=N82E16819103996

Even at it's MSRP of $115, it still isn't worth it. There are other better performing options from both AMD and Intel at that price point.

 

[pjmelect]

After studying the benchmarks for a while I realized that in percentage terms there is not much to chose between any of them. After all you are unlikely to notice a 20% difference between two computers. Also if AMD's prediction that they will increase the speed by 15% every year is true, then in two years time they will have the fastest CPU. Of course Intel will also increase the speed of their CPU's in the same time. In practical terms there is not much to chose between any of them except price.

 

[fazers_on_stun]

And yet the 8120 and 8150 are both sold out or out of stock on Newegg.

I've heard that's because they didn't get many. Dunno how true that is, however. If it is, my guess is that AMD did something of a paper launch with desktop BD, reserving most of them for the much more lucrative server market. Anyway, supply will probably remain limited until GF gets their yields up.

 

[mailpranshu]

Sorry guys...

I am an AMD fanboy myself and have been a rocking an old Athlon 5600+. I resisted the temptation to pick up a i7 last year.

Now I feel stupid waiting so long.

I don't care if Intel is bribing or whatever if they are giving us better products all the time. There is no David v/s Goliath situation in a market like this. All that matters is ingenuity not only money( Ford v/s Suzuki(

I am ordering a 2600K.

Thank you AMD... :-(

 

[amk-aka-Phantom]

Get real Tom's. It's a fact that Windows 7 doesn't even recognize BD's architecture, which is why is does so bad in the benchmarks. Windows 7 doesn't know where to place the threads and they end up out of order, hence the bad benchmarks. It's a good processor, but today's software not only does not recognize it's architecture but can't utilize it efficiently. Why not show some benchmarks from the server side, then you will see it shine and put a hurtin' on sandy bridge zeons.

No, you get real. THIS is AMD's failure - they couldn't design something that works well with the current software. I don't see why everyone should go and blame the software that works great with Intel CPUs instead of blaming the poorly-designed Bulldozer.

 

[shinkueagle]

I USED to be a die-hard AMD Fanboy.... Now I have forgot what AMD stands for!

Is it:
---> ALWAYS MISERABLY DISFUNCTIONAL

I think that SHOULD be just about right....

 

[tanjo]

Really disappointing. They should've named it "Bulldozed"

 

[azxcvbnm321]

This is really sad. I bet they knew it was a bust which is why they delayed it for so long, attempting to fix the problems, but the problems are with the entire design and couldn't be fixed.

AMD just doesn't have the resources and lately doesn't have the same quality of management as Intel. They probably knew 1 year ago that this was not going to be a successful design, but I bet that they had put so much hope and R&D into Bulldozer that no one had the guts to kill off the entire design and start fresh. Management probably understood that scrapping the whole thing would put them...where they are now, hopelessly out of competition. Only they are a year behind where they would be if they had scrapped the entire Bulldozer design.

Basically AMD made the same mistake Intel did with Pentium 4, but Intel has the resources to bounce back. If you're optimistic, you'll hope that AMD realized early that Bulldozer is a dead end and has been working on a completely new design. Bulldozer, like Pentium 4, is just a stopgap until the new design can be put into production. But then why announce Piledriver? If this optimistic view sounds like fantasy, well it probably is. AMD execs are just too incompetent and afraid to admit what we all know, Bulldozer is DOA. When you've put all your hopes and resources into a project, it takes real guts and leadership to admit Bulldozer stinks and to order the research department to start fresh. I don't think AMD has that kind of quality management. It's over folks, at least for another 2 years, Intel should completely dominate whatever space they CHOOSE to. Yes, it's Intel's choice whether to put AMD out of business by pricing a better i3 at slightly lower prices than a similar Bulldozer, or go for profits and let AMD barely break even with their best CPU.

 

[silverblue]

Piledriver is a revision of Bulldozer. It's not as if AMD are scrapping the architecture. If Bulldozer is such a failure and a bad idea, they wouldn't be considering three more generations based upon it. Unless, of course, somebody in management had a serious screw loose.

 

[yankeeDDL]

Ouch!

 

[elme]

I'm under the impression that Intel's motherboards are generally very efficient compared to other manufacturers'. So it would be interesting to see 1155-platform results with a top-of-the-range Asus motherboard, too.

 

[pelov]

[citation][nom]silverblue[/nom]Piledriver is a revision of Bulldozer. It's not as if AMD are scrapping the architecture. If Bulldozer is such a failure and a bad idea, they wouldn't be considering three more generations based upon it. Unless, of course, somebody in management had a serious screw loose.[/citation]

The odd thing about Bulldozer is that if they keep the same architecture as they have today and Intel were to stall but software kept evolving then the AMD chips would pull ahead of Sandy Bridge. Of course, this won't happen. Unfortunately, I think they're stuck between a rock and a hard place. Changing their architecture and core design now would mean they'll probably have to slowly revert to something with more cores, but if they don't significantly change things then people will quit buying them and they'll lose out on even more revenue (Bulldozers are sold out not because they're awesome, but because GloFo has some serious issues with yields). Let's be honest, why would someone pay more for a product that's inferior and uses too much power compared to the alternative? As soon as the hype settles down and people pick up on the fact that these things are netburst power suckers the sales will quickly dry up, too.

I think either way, they really need to change certain things. The power consumption is far too high for a 32nm chip and 2 billion transistors is unnecessary. If they don't change things they'll lose even more market share to Intel on server and desktop, but a mere clock speed increase just won't cut it here, that would only increase the power consumption figures. Remember, the chips overclock well, the headroom isn't the problem. The issue is that you need to overclock them so high to keep them competitive,pushing wattage through the roof. that's not a feature of modular design but a design flaw and something they had to overlook in order to feed us 8 cores.

 

[silverblue]

I have a few thoughts about Bulldozer (my own opinions, and not necessarily right)...

1) AMD releaseda server chip on the desktop. Somebody has to decide on a desktop variant without all the complicated I/O and cache architecture associated with a server CPU.
2) It'd make more sense to power down cores on their own as opposed to a modular level.
3) The L1 cache is too small and there's obviously a problem with it as it stands (OS patch?).
4) The cache overall is far too slow.

I don't think the idea of Bulldozer is bad, just the initial execution. I'm wondering if a faster uncore would help. Is the front end too constrictive? Did AMD just want the first 8-core desktop CPU as opposed to one that works as it should and performs like an 8-core?

 

[pelov]

Well, the pipelines are far too long and the IPC suffers for it. The cache problem is rather well-known now and the speeds actually decreased from Phenom II to Bulldozer, and if you throw Intel into the mix it becomes embarrassing.

It does look like a server CPU, doesn't it? Far more than a desktop chip, for sure. The clock speeds are a result of the longer pipeline they have and a last-ditch attempt to push the IPC up. Like I said, though, the chip itself has 2 billion transistors 8 cores (some threads are actually shared), and since the power consumption is already far too high the notion that they'll actually increase the clock speeds for future chips or even Piledriver without a serious restructuring of the architecture will not bode well for them... at all.

In a server environment they can potentially downclock these things and get their TDP low enough that they'll be attractive if they're priced correctly. On the desktop, however, downclocking or even increasing clock speed will prove a fool's attempt to get the IPC up.

 

[LORD_ORION]

[citation][nom]mailpranshu[/nom]Sorry guys...I am an AMD fanboy myself and have been a rocking an old Athlon 5600+. I resisted the temptation to pick up a i7 last year. Now I feel stupid waiting so long.I don't care if Intel is bribing or whatever if they are giving us better products all the time. There is no David v/s Goliath situation in a market like this. All that matters is ingenuity not only money( Ford v/s Suzuki(I am ordering a 2600K.Thank you AMD... :-([/citation]

Well, now you should wait for Ivy Bridge

 

[shinkueagle]

AMD BULLDOZER IPC improvement....

AMD ---> ABSURDLY MORONIC DEVICES

IPC ---> IRRELEVANT PROCESSING CHIP

 

[eddieroolz]

Gosh, I had hope that Zambezi would at least be able to beat its predecessors in this test. There were hints in the opening paragraphs that this was not going to happen, however. And my prediction came true.

It seems as if something seriously went wrong during the development of the Zambezi - it was as if the engineers failed to test the pre-production hardware on benchmarks. I can't for the life of me understand how engineers did not catch this processor using more power than its predecessors all the while it's taking longer to do a task.

 

[ohim]

[citation][nom]compton[/nom]Geez, the 2700K is creeping up on $400. Thanks a lot AMD. You're off my Christmas list.[/citation]
Can i say now HAHA to the intel fans who think they can go without amd ?

 

[shompa]

AMD people needs to stop harping about the server market. X86 have grown extremely in the server market taken about 60% of the server market. AMD only have 5-6% of the X86 server market.

So If AMD is binning server CPUs, they do it for a really small market.

The amazing thing is that AMD is making a profit.

Prediction 1-5 years:
Low end: ARM
Middle end ARM/Intel/AMD
High end: Intel
Server: Intel/ARM64/GPU

AMD will be gone before 5 year. Maybe someone will buy the company. Their market cap is 4 billion, so its not to expensive. The problem is that Intel have to approve the transfer of the X86 license.

(people should read a little history. AMD was had 2 excellent series: K6/K7. Both of their performance is nothing that AMD had to do with. K6 was a NextGen design. K7 used a lot of Alpha things. Today there is no company that AMD can borrow technology from. Therefore AMD will never be as fast as Intel. Global foundries is also a mess. Biggest mistake in AMDs history to sell its fabs)

 

[opmopadop]

Sry if I doubled up on somebody elses comment... Tom, any chance we can see a Windows 7 to Windows 8 power comparison seems Win8 runs these chips differently (parked cores etc)?

 

[Why_Me]

AMD is more or less a total fail in the desktop cpu market. I wouldn't use one for a paperweight if it was given to me.

 

[_Pez_]

AMD sucked it big time with this for me this is a non 8 core CPU instead it is 4 core with two treads per core. that it is for me this cpu and amd failed claiming it is a 8 pseudocores. I would say that those are almost a core.

 

[Why_Me]

AMD sucked it big time with this for me this is a non 8 core CPU instead it is 4 core with two treads per core. that it is for me this cpu and amd failed claiming it is a 8 pseudocores. I would say that those are almost a core.

It could have a 144 cores...who cares? A piece of crap is a piece of crap. People make too much out of the amount a cores a cpu has.