AMD FX: Energy Efficiency Compared To Eight Other CPUs

[memadmax]

Just admit it BD is a spectacular bust.... if you compare it to todays procs.....

But if you go back 2 years, this would have been a great processor to buy and would have given intel a run for its money... but...

It's too late... turn it around before christmas(literally) or AMD is gonna have a serious setback.

 

[dgingeri]

[citation][nom]memadmax[/nom]Just admit it BD is a spectacular bust.... if you compare it to todays procs..... But if you go back 2 years, this would have been a great processor to buy and would have given intel a run for its money... but...It's too late... turn it around before christmas(literally) or AMD is gonna have a serious setback.[/citation]

Not entirely. It works well in servers as a file server. Those partial dual core modules work well for heavy multi-threaded integer loads. It works well for file servers, backup servers, and virtualization hosts for the previous two. Sure, it's a bust for desktop or workstation use, but that doesn't mean it's a total bust.

 

[memadmax]

[citation][nom]dgingeri[/nom]Not entirely. It works well in servers as a file server. Those partial dual core modules work well for heavy multi-threaded integer loads. It works well for file servers, backup servers, and virtualization hosts for the previous two. Sure, it's a bust for desktop or workstation use, but that doesn't mean it's a total bust.[/citation]

I didn't say a total bust, you have your points but that is not what this proc was meant for right? This proc was meant to be a consumer grade and apparently *was* intended for low power apps(in actual power usage) Like I said, if it was released two/three years earlier, it would put AMD in the same curve of intel, unfortunately it was not.....

Long live, AMD.

 

[pelov]

[citation][nom]dgingeri[/nom]Not entirely. It works well in servers as a file server. Those partial dual core modules work well for heavy multi-threaded integer loads. It works well for file servers, backup servers, and virtualization hosts for the previous two. Sure, it's a bust for desktop or workstation use, but that doesn't mean it's a total bust.[/citation]

The #1 priority in a server environment is low heat/power draw. Bulldozer doesn't do well here. If the chips were underclocked they may be ok file server parts, even file server/backup servers. But if you're just going to undervolt and underclock a chip then why not buy a 2600k for the same task? hell, even a 2500k.

I think the only viable reason to buy one would be if you're running a very select set of programs, integer-based and steering away from anything FPU related. only then it could potentially pay off. This, though, is not exactly the segment of the consumer market AMD was looking to attact with this chip.

I've been a die-hard AMD fan for ages and even I can't find a single reason to come here and defend this chip. It really is just that bad. Hopefully my unlocked and bumped 555 will serve me for another 2 generations or whenever they're ready to completely abandon the bulldozer architecture. But judging by what they're saying about piledriver and trinity, I'm just not sure things will get better.

 

[wiyosaya]

I DO consider myself an AMD fanboy.

However, I bet that the FX-8150 and any CPU that AMD releases that performs as dismally as the 8150 will not be in the market for long just like similar fiascos from Intel such as "Core".

I have to wonder if the release of the 8150 by AMD was just to satisfy the fact that they said that they were going to release BD, and decided to do so even though they likely know they have a crap chip. I would not be surprised if the recent high-level management changes were a result of the board being informed of the impending fiasco named Bulldozer.

So, being an AMD fanboy, I'll hope that they do address the IPC and power problems soon and put the fiasco named Bulldozer into the dump.

In the meantime, I will not pay arms and legs to Intelopoly.

 

[majorlag]

Yet another article were the tests do not use the DDR3 RAM that Bulldozer is speced for. Yet they gladly put in triple channel RAM in for the Intel system, but starve the BD of its 1866 DDR3 ram support.

Can we please get a decent review that shows BD with DDR3 1866. The last article showed one test that showed 1866 ram being faster, and yet they don't bench all the tests. Yes it is faster by seconds, but since some tests are only 30second spread, every second counts.

I don't care if BD is first, but I am sure with the right RAM it will do way better then all of these reviews keep showing.

 

[wiyosaya]

[citation][nom]amk-aka-phantom[/nom]AMD fanboys are so hilarious. Just ignore all benchmarks where Bulldozer loses... Core 2 Quad Extreme QX9770 was an extremely powerful CPU targeted at multimedia production; I can easily believe it beating Bulldozer in certain tasks. It is not an error.[/citation]
Re-read the post you slammed. It explicitly said the LATTER, meaning the Core2 Extreme QX9770, wins. And you probably wonder why people hammer you.

 

[nforce4max]

I was a fanboy of sorts about amd until BD came out, slower, hotter, and well crappier. AMD was great bang for buck on the cheap compared to Intel where one has to pay more to get the same results but after BD came out that has changed. Now older Phenom and low end Intel cpus are the best bang for buck. I wouldn't pay $200+ for a BD let alone $100. Instead from now on I am totally ignoring BD as a product in future builds and siding with cheap i3 and i5 along with phenom.

At least I am able to recover some value out of my AM3+ board with a cheap p2 x4 820 @ 3.5ghz and using two cheap gtx 280 in sli.

 

[DoomsWord89]

Oh my, 209W for a stock 8150?.....Madness!

 

[gonchuki]

I wonder how great could BD have been had AMD used a proper size L1 cache and get all 3 caches to a lower latency. Ever since Brisbane went into production they have been getting worse and worse cache latencies, up to the ridiculous situation where they are now.

They need to launch a FX-8155 *FAST* with at least this issue fixed. It worked for the P4 where they had a similar deep-pipeline+crappy-cache design, it's most probable that it will have the same effect here.

 

[pelov]

[citation][nom]majorlag[/nom]Yet another article were the tests do not use the DDR3 RAM that Bulldozer is speced for. Yet they gladly put in triple channel RAM in for the Intel system, but starve the BD of its 1866 DDR3 ram support.Can we please get a decent review that shows BD with DDR3 1866. The last article showed one test that showed 1866 ram being faster, and yet they don't bench all the tests. Yes it is faster by seconds, but since some tests are only 30second spread, every second counts.I don't care if BD is first, but I am sure with the right RAM it will do way better then all of these reviews keep showing.[/citation]

The tests are done at that speed because the phenom II can only support up to ddr3 1333 mhz. If the Phenom II isn't tested and the BD goes up against Sandy then you use the same ram and higher than 1333 (1866, 1600, whatever). Either way, I can assure you from benchmarks I've seen that the ram isn't the issue here and isn't holding the chip back. I've heard this claim time and time again and every single time the Phenom II was the limiting CPU in terms of which ram to choose.

 

[LORD_ORION]

Hrmmmm

What would be interesting to see (performance and power tests)
1) Phenom1 vs Phenom II
2) Phenom1 vs 775 compared to PhenomII vs 1155

Why? To predict if Bulldozer has a more optimistic future then the last new architecture?

 

[AMD fanboy]

ofc FX-8150 consume more power, especially under full load
repeat this 100 times: 8 core FX vs 4 core SB

control consumption when runing every single benchmark and tell us how much it consumes and how much it gets under full load

do a ubuntu linux benchmarking 2, one is already done:
http://www.phoronix.com/scan.php?page=article&item=amd_fx8150_bulldozer&num=1

and tell us how much intel pays you?

 

[pelov]

[citation][nom]amd fanboy[/nom]ofc FX-8150 consume more power, especially under full loadrepeat this 100 times: 8 core FX vs 4 core SBcontrol consumption when runing every single benchmark and tell us how much it consumes and how much it gets under full loaddo a ubuntu linux benchmarking 2, one is already done:http://www.phoronix.com/scan.php?p [...] ozer&num=1and tell us how much intel pays you?[/citation]

if you're going to link phoronix's review then you may also want to state that they too were disappointed with the processors performance and were pretty unsure what to say. On a linux workstation it's not bad because so many linux programs are so highly threaded, but linux only represents 2% of the desktop market and the same can't be said about a regular windows platform. To boot, phoronix never even included the power consumption figures.

 

[AMD fanboy]

if you're going to link phoronix's review then you may also want to state that they too were disappointed with the processors performance and were pretty unsure what to say. On a linux workstation it's not bad because so many linux programs are so highly threaded, but linux only represents 2% of the desktop market and the same can't be said about a regular windows platform. To boot, phoronix never even included the power consumption figures.

and what is the future?
highly threaded programs maybe?

 

[pelov]

[citation][nom]AMD fanboy[/nom]and what is the future?highly threaded programs maybe?[/citation]

I agree wholeheartedly. It absolutely is. Unfortunately, that future is still a few years away and requiring that you buy a new operating system to fix the scheduling issues associated with the bulldozer's design isn't exactly a selling point. Have you seen the threads regarding the BSOD issues and crashing issues certain people are having that's related to the new chip? You shouldn't be relying on a future operating system in order to show a 10% performance increase, never mind the stability factor.

They released a product that's too early for its time. The blame here isn't on the software developers, because getting the entire industry to suddenly start doing more work and optimizing their programs/games for 2+ or 4+ cores isn't how you win anybody over. AMD just didn't read the industry correctly and offered us something we don't need and something that nobody who has a decent amount of hardware/software knowledge would even want. And then they made it consume a butt-ton of power

 

[silverblue]

[citation][nom]wiyosaya[/nom]Re-read the post you slammed. It explicitly said the LATTER, meaning the Core2 Extreme QX9770, wins. And you probably wonder why people hammer you.[/citation]
I think he was insulted by my assertation that Bulldozer should always beat the QX9770 regardless of synthetics that heavily favour Intel CPUs. To be honest, I shouldn't have drawn attention to it.

 

[AMD fanboy]

I agree wholeheartedly. It absolutely is. Unfortunately, that future is still a few years away and requiring that you buy a new operating system to fix the scheduling issues associated with the bulldozer's design isn't exactly a selling point. Have you seen the threads regarding the BSOD issues and crashing issues certain people are having that's related to the new chip? You shouldn't be relying on a future operating system in order to show a 10% performance increase, never mind the stability factor.

They released a product that's too early for its time. The blame here isn't on the software developers, because getting the entire industry to suddenly start doing more work and optimizing their programs/games for 2+ or 4+ cores isn't how you win anybody over. AMD just didn't read the industry correctly and offered us something we don't need and something that nobody who has a decent amount of hardware/software knowledge would even want. And then they made it consume a butt-ton of power

so you are suggesting new CPU should not look to the future? roflmao?!

ofc it has issues, its new architecture, thats normal

multithread software are present now, just not so much for home desktop multimedia/gaming pcs
OS improvement should bring out more from this CPUs

how much power bulldozer consumes? tell me pls? it consumes more power under full load, but how often is this processor under full load?
the real question is how much power it consumes comparing vs SB performing the same task(real task, not the full load bench task)

 

[pelov]

I don't doubt multi-threading is the future, but arguing for someone to buy one of these doesn't make sense either. You design a chip that suits the market and provides enough room for advancement until your next chip is released. They didn't do that. They threw IPC out the window and assumed all software you'd run will be multi-threaded. As it turns out, a majority of software you run can't even utilize more than 2 cores, nevermind 8.

You can't even buy one of these and wait 5-10 years, then take it out and fire it up and have it perform well with chips that are 5-10 years in the future, because they, too, will outperform this thing.

Sandy Bridge had issues too, but that was SATA 6GB/s related and nothing to do with the CPU. Try firing up Portal 2 on your new bulldozer and see what happens. Then ask yourself, when was the last time a CPU actually froze and crashed your game because it was "so far ahead of its time." The answer is never and it shouldn't ever cause such issues. You need to account for backwards compatibility on the desktop and with that there should be a good amount of focus on IPC. Instead they threw it out the window and your Phenom II is generally faster than a bulldozer at your average task.

Performance-per-watt is what I assume you're getting at, but there too it fails. I'm not going to be running Blender 24/7, therefore this thing has no use to me or the other 99.5% of us.

 

[Inferno1217]

Comparing 4 and 6 cores to 8 cores is in fact quite unfair. Actually if you add cores to intel or take away cores from AMD they seem to be on par with each other with Intel and AMD both losing and winning some efficiency tests and not by much. Must we not forget AMD has 8 physical cores vs Intels 4-6 cores and imaginary cores. Both of my systems are intel and the HT they have is mostly a gimmick. Most people on here recommend 2500k over the 2600k as its not really worth the money to most.

Example 2500K double the cores (4) to 8 and double the Power consumption (22) for multi-threaded apps to 44 and Intel and AMD are the same. With that said AMD still needs to fix the CPU and get it more on par with SB and IB. I would still buy a 2500k over the AMD platform unless we are talking about Llano.

 

[AMD fanboy]

You need to account for backwards compatibility on the desktop and with that there should be a good amount of focus on IPC. Instead they threw it out the window and your Phenom II is generally faster than a bulldozer at your average task.

how it performs in multitasking?
all of us is multitasking all the day, all the way

Performance-per-watt is what I assume you're getting at, but there too it fails. I'm not going to be running Blender 24/7, therefore this thing has no use to me or the other 99.5% of us.

link pls for performance per watt
on tomshardware performance per core(for single threaded apps) is shown and when you disable 4 cores in fx 8150 performance is better(for single threaded apps) and power consumption is smaller

from tomshardware review performance per core:
all running our single-threaded...

 

[pelov]

The performance-per-watt: http://www.tomshardware.com/reviews/fx-power-consumption-efficiency,3060-12.html

 

[pelov]

tom's you cut me off!

http://www.tomshardware.com/reviews/fx-power-consumption-efficiency,3060-13.html

multithreaded ^^.

You can't disable the cores or have the scheduler sit them in idle and use the others, because windows 7 doesn't do that. Meaning you'll have to add $150 price tag to the bulldozer CPU you're looking to buy if you want to get about a 10% performance increase and a 10-20% power consumption decrease (which still puts it behind intel and even behind its older chips. Also, windows 8 won't be here till next year).

"Multi-tasking" isn't where it excels, either. That's actually where it can potentially fall apart. Throw in a single FP intensive task in that group of programs you're running and your system has damn near completely stalled.

Sorry, bud. I love AMD as much as you do, unfortunately what they did here was disappoint all of us

 

[AMD fanboy]

tom's you cut me off!

http://www.tomshardware.com/reviews/fx-power-consumption-efficiency,3060-13.html

multithreaded ^^.

You can't disable the cores or have the scheduler sit them in idle and use the others, because windows 7 doesn't do that. Meaning you'll have to add $150 price tag to the bulldozer CPU you're looking to buy if you want to get about a 10% performance increase and a 10-20% power consumption decrease (which still puts it behind intel and even behind its older chips. Also, windows 8 won't be here till next year).

but that is not performance per watt
do performance per watt for multi-threaded apps where 8150 does great

"Multi-tasking" isn't where it excels, either. That's actually where it can potentially fall apart. Throw in a single FP intensive task in that group of programs you're running and your system has damn near completely stalled.

Sorry, bud. I love AMD as much as you do, unfortunately what they did here was disappoint all of us

so many reviews and so different tests for multitasking
heres 1 rating fx 8150 with 3/5 stars

http://www.expertreviews.co.uk/processors/1287799/amd-fx-8150/2

Under incredibly intensive multitasking, the FX-8150 has the edge over Intel's Core i5-2500K

 

[Area51]

[citation][nom]AMD fanboy[/nom]so you are suggesting new CPU should not look to the future? roflmao?!ofc it has issues, its new architecture, thats normalmultithread software are present now, just not so much for home desktop multimedia/gaming pcsOS improvement should bring out more from this CPUshow much power bulldozer consumes? tell me pls? it consumes more power under full load, but how often is this processor under full load?the real question is how much power it consumes comparing vs SB performing the same task(real task, not the full load bench task)[/citation]
You sir are an Idiot. At this point AMD sucks at any workload.
lets just wait and see how the server version "If it ever becomes available" will do against the E5 xeon CPU's from Intel, then try to defend them.