Do AMD cpus at least give a smoother desktop experience w/more cores?

[AtenRa]

Nobody that use AMD and efficient in the same sentence can be taken seriously.

You should see my SFF A8-7600.

 

[Madpacket]

I have an MSi 970 gaming and a FX8320e overclocked to 4.3ghz with a modified/OC R9 290. i'm into playing Warthunder and got back into BFBC2 online gaming. My set-up does fine playing at 1920x1080 everything cranked on a Samsung 23"monitor I got from Costco several years ago. I bought the CPU off ebay on a buy now, and the MSi 970 off of another forum. Both were cheap and "new condition" I know exactly why. All the benches showed up, the FX did'nt do so well and the sellers got pushed by hype into buying more expensive intel set-ups.

Works for me. Paired with the r9 290 which was a previous bitcoin miner I bought locally off Craiglist for $175 it all works very well.

Good for you for not falling for the hype. It's amazing what kind of deals you can acquire due to peoples misconceptions or preconceived notions that's something should suck because they were told so.

 

[MiddleOfTheRoad]

Nobody that use AMD and efficient in the same sentence can be taken seriously.

AMD's Radeon Nano is a pretty compelling product when it comes to efficiency. Let's face it, nothing can touch Carizzo's efficiency at its respective node.

Is the AMD FX efficient? Yeah, probably not so much.

 

[frozentundra123456]

AMD's Radeon Nano is a pretty compelling product when it comes to efficiency. Let's face it, nothing can touch Carizzo's efficiency at its respective node.

Is the AMD FX efficient? Yeah, probably not so much.

Yea, but the point is, Carizzo is on an outdated node, so it is kind of irrelevant. It still amazes me that posters use AMD's node disadvantage to excuse their performance. What matters is the performance of the final product. It is a major part of the job to field a competitive node, which AMD has failed miserably at for several years.

 

[AtenRa]

Yea, but the point is, Carizzo is on an outdated node, so it is kind of irrelevant. It still amazes me that posters use AMD's node disadvantage to excuse their performance. What matters is the performance of the final product. It is a major part of the job to field a competitive node, which AMD has failed miserably at for several years.

Compare the performance and perf/watt of 22nm FF Core i3 3225 vs 28nm planar 55W A8-7600

http://forums.anandtech.com/showthread.php?t=2420502

The AMD A8-7600 is not only faster, it also has higher perf/watt with a node disadvantage. This is simple amazing, if Intel would do that everybody would talk about the second coming of Jesus.

In iGPU there is no competition, even at lower TDP of 45W, the 28nm PLANAR Kaveri is dominating every single benchmark.

 

[frozentundra123456]

Standard operating procedure: cherry picked benchmarks comparing AMD to a 3 generation old Intel product.

 

[AtenRa]

Standard operating procedure: cherry picked benchmarks comparing AMD to a 3 generation old Intel product.

1. I have included the url of the thread, you can go there and see all the benchmarks. You didnt believe i was going to post every single benchmark again did you ??

2. Kaveri is also two years old, but i havent seen anyone indicating this when compared against SkyLake.

3. Instead of acknowledging that even with a huge node disadvantage AMD Kaveri was able to outperform especially in perf/watt an Intel 22nm FF CPU you resorted to excuses.

 

[Yuriman]

I think his point is that you can't really buy an i3 3xxx - it has been replaced. We'd be comparing Ivy Bridge's successor to Kaveri's successor if there was one.

Why the i3 3xxx anyway? The 4xxx is also 22nm, is similarly priced, and can still be purchased, though I'd question the wisdom of anyone who does that unless they find a compelling sale, given that Skylake i3's are not significantly more.

EDIT: And I'm seeing 25% less total system power consumption for the 22nm i3 4xxx when compared with 55w Kaveri chips, while providing better performance.

EDIT2: It's certainly true that AMD has a more efficient iGPU though, and that has value in some use-cases. At least, I think it's still more efficient than Skylake's?

 

[USER8000]

Standard operating procedure: cherry picked benchmarks comparing AMD to a 3 generation old Intel product.

Wait did I just miss Broadwell Core i3 for desktop?? I think you might find Haswell Core i3 chips were common in the channel and in prebuilt PCs than the Skylake ones last year,when he did the review,and Haswell was the last generation. So not three generations for desktop - one generation and even then Skylake Core i3 chips were only available in a trickle in most UK retailers until December 2015 onwards.

But more importantly,what CPUs have you used or owned in the last few years??

I have had or use almost everything upto Haswell Core i7 chips - so any chance you can post your counter benchmarks which you run yourself? Would be interesting to see what your findings are.

I see a lot of people being critical of somebody who has taken a lot of time to run benchmarks for the community and people just whining at him.

I don't see anything biased about the benchmarks - the software he uses is found in many reviews from websites too.

Plus you can send him a Core i3 6100 to benchmark at you own cost,instead of handing out accusations.

I am sure he will add one at a later date when he can,but hardware is not free,and neither is time for benchmarking.

You might want to try it yourself one day - its why I don't complain at Anandtech if their reviews take time. Benchmarking takes time.

 

[2is]

Wait did I just miss Broadwell Core i3 for desktop?? I think you might find Haswell Core i3 chips were common in the channel and in prebuilt PCs than the Skylake ones last year,when he did the review,and Haswell was the last generation. So not three generations for desktop - one generation and even then Skylake Core i3 chips were only available in a trickle in most UK retailers until December 2015 onwards.

But more importantly,what CPUs have you used or owned in the last few years??

I have had or use almost everything upto Haswell Core i7 chips - so any chance you can post your counter benchmarks which you run yourself? Would be interesting to see what your findings are.

I see a lot of people being critical of somebody who has taken a lot of time to run benchmarks for the community and people just whining at him.

I don't see anything biased about the benchmarks - the software he uses is found in many reviews from websites too.

Plus you can send him a Core i3 6100 to benchmark at you own cost,instead of handing out accusations.

I am sure he will add one at a later date when he can,but hardware is not free,and neither is time for benchmarking.

You might want to try it yourself one day - its why I don't complain at Anandtech if their reviews take time. Benchmarking takes time.

I think what you missed is the point, actually. What is it that you're even talking about?

No one even bothers posting Intel benchmarks anymore, you now why? Because this silliness that AMD can compete has been busted 5 years ago and they haven't released anything new since then while Intel has. Try and keep up.

 

[USER8000]

Speaking of the devil,looks like he benchmarked the Core i3 6300:
http://forums.anandtech.com/showthread.php?t=2456159&highlight=

So he actually already benchmarked the top end Core i3 and the first reply was from the same poster,who called him biased. Emm,WTF??

 

[USER8000]

I think what you missed is the point, actually. What is it that you're even talking about?

No one even bothers posting Intel benchmarks anymore, you now why? Because this silliness that AMD can compete has been busted 5 years ago and they haven't released anything new since then while Intel has. Try and keep up.

Because,he is whining about people being biased,and this is an enthusiast forum,and some of those comparisons made were interesting to read since he includes a whole range of chips - websites only really care mostly about higher end chips.

It's hilarious,when the P4 was beaten by the A64 or XP in many benchmarks at lower power consumption,using your logic nobody should have bothered posting P4 benchmarks...

Nobody should bother having benchmarks outside an overclocked GTX980TI since it conqueurs all other graphics cards.

I think some of you have missed the whole point of being an enthusiast at times - seems E-PENIS is more important.

You might want to go back to what this mod said earlier in the thread:

And no one should. That is not what the culture was/is about. We shared/stealth bragged about how much we could get over stock, and the performance improvement that went with it. We shared which chips from which fabs from which batch were the best. What could you do on a tight budget with it. What could you do if you were willing to throw money at it. From ultra budget board to the highest end, and which settings to use. Physical mods etc. etc.

I still see some of it, but all this versus nonsense has taken over. It is like high school kids of the past, sitting across from each other shouting which group has more spirit.

Hyperbole, obfuscation, talking points, an overclocker craves not these things. :awe:

Instead of complaining at the poor bloke who is making some effort at benchmarking chips on an "enthusiast computer" forum,why don't some of you then buy the same chips and do "better" benchmarks??

 

[SlowSpyder]

Nobody that use AMD and efficient in the same sentence can be taken seriously.

I can underclock my FX pretty well. That's a 10% higher base than an FX8350, at a good chunk less voltage than a factory set FX8350. Not bad.

 

[2is]

Because,he is whining about people being biased,and this is an enthusiast forum,and some of those comparisons made were interesting to read since he includes a whole range of chips - websites only really care mostly about higher end chips.

It's hilarious,when the P4 was beaten by the A64 or XP in many benchmarks at lower power consumption,using your logic nobody should have bothered posting P4 benchmarks...

Nobody should bother having benchmarks outside an overclocked GTX980TI since it conqueurs all other graphics cards.

I think some of you have missed the whole point of being an enthusiast at times - seems E-PENIS is more important.

You might want to go back to what this mod said earlier in the thread:



Instead of complaining at the poor bloke who is making some effort at benchmarking chips on an "enthusiast computer" forum,why don't some of you then buy the same chips and do "better" benchmarks??

When the P4 was getting it's ass kicked, at least Intel was still releasing new versions of the P4 on a regular basis so it made sense to continue to compere them. That's not the same thing as AMD's FX line which hasn't changed in years.

Like I said, try and keep up. And for future reference, don't pretend to be a logic expert when you can't even apply it. Like comparing a product line which had regular updates to one that hasn't, as you just did here.

 

[2is]

I can underclock my FX pretty well. That's a 10% higher base than an FX8350, at a good chunk less voltage than a factory set FX8350. Not bad.

Being more efficient then an inefficient processor isn't saying much. And it's saying even less when the same process can be done to processors that are already far more efficient.

 

[Yuriman]

I can underclock my FX pretty well. That's a 10% higher base than an FX8350, at a good chunk less voltage than a factory set FX8350. Not bad.

Using your Cinebench results, I decided to compare against my own system, as listed below in my signature. I used a Kil-A-Watt and measured from the wall for my power numbers, and used CPU-Z to obtain my voltages.

Total system power
FX @ 4.4ghz, 1.2250v
Idle: 120w
Cinebench: 242w
Score: 688

Ivy Bridge i5 @ 3.6ghz, 1.016v (stock speed, undervolted)
Idle: 37.4w (+/- 0.1w)
Cinebench: 69.0w (+/- 0.3w)
Score: 505

~

When comparing total system power consumption, my i5 comes out having approximately 2.6x the performance per watt.

It's not just the FX platform's high parasitic power draw skewing the results though. When you subtract idle watts from load watts (effectively isolating the CPU's power draw), and compare:

i5 Delta Watts: 31.6w
FX Delta Watts: 122w

The 3 generation old i5, in this light, has approximately 2.8x the performance per watt. Cranking it up to 4.0ghz reduces performance per watt slightly, but it still delivers 2.4x the performance per watt.

I highly doubt it's directly correlative, but AtenRa's testing with i3's shows Haswell as having ~33% better performance per watt than Ivy Bridge, and Skylake as having ~60% better performance per watt than Ivy Bridge. (watt hours to complete Cinebench R15 x264 benchmark). Also, I'm under the impression that HT CPUs have slightly better performance per watt than non-HT CPUs, which would make the FX look even worse when compared with an i7 (or i3, though they're not in the same ballpark in terms of performance).

Now, admittedly, your FX system still scores 36% higher in the benchmark, which has value in and of itself, but the power numbers don't look good at all.

FWIW

 

[AtenRa]

I think his point is that you can't really buy an i3 3xxx - it has been replaced. We'd be comparing Ivy Bridge's successor to Kaveri's successor if there was one.

Why the i3 3xxx anyway? The 4xxx is also 22nm, is similarly priced, and can still be purchased, though I'd question the wisdom of anyone who does that unless they find a compelling sale, given that Skylake i3's are not significantly more.

EDIT: And I'm seeing 25% less total system power consumption for the 22nm i3 4xxx when compared with 55w Kaveri chips, while providing better performance.

EDIT2: It's certainly true that AMD has a more efficient iGPU though, and that has value in some use-cases. At least, I think it's still more efficient than Skylake's?

The comparison to the Core i3 Ivy was only to illustrate that even with a HUGE node advantage of FineFets, and still the AMD Kaveri on a inferior process on 28nm PLANAR was still faster and with better perf/watt. Even in CPU workloads, not to mention iGPU performance. Ohh, dont forget at the same TDP of 55W.

 

[AtenRa]

but AtenRa's testing with i3's shows Haswell as having ~33% better performance per watt than Ivy Bridge, and Skylake as having ~60% better performance per watt than Ivy Bridge. (watt hours to complete Cinebench R15 benchmark).

You mean x264 ?? because i havent measured wh on Cinebench

 

[Yuriman]

^ I'd like to add that I'm using a $25 power supply (which is still almost twice as powerful as is necessary), a single 120mm case fan, and this cooler:

^ Temps during the benchmark never exceeded 57c, the CPU fan never went above 900RPM, and the case fan not above 600RPM, remaining effectively silent.

The stock cooler would've been more than adequate, even for an 8-900mhz OC, which is why I hold to the idea that AMD CPUs necessitate spending more on cooling and power supplies (needs to handle an extra ~200w minimum), which in many cases hurts their value proposition.

You mean x264 ?? because i havent measured wh on Cinebench

My apologies, I got mixed up. Yes, I had been looking at x264.

 

[Yuriman]

I'm sure you know this (as you've accounted for it in your benchmarking), but I'd like to point out for the benefit of others that comparing TDP isn't always very useful. My i5 for instance, completely stock, has only a ~55w power delta at the wall under linpack/AVX torture, despite having a 77w TDP. Most CPUs will exceed or be well under their rated TDPs in various load situations.

 

[AtenRa]

I'm sure you know this (as you've accounted for it in your benchmarking), but I'd like to point out for the benefit of others that comparing TDP isn't always very useful. My i5 for instance, completely stock, has only a ~55w power delta at the wall under linpack/AVX torture, despite having a 77w TDP. Most CPUs will exceed or be well under their rated TDPs in various load situations.

Im not talking about power consumption but thermals and how that will effect Turbo operation and performance of the CPU. The 22nm FF node has a HUGE advantage at lower TDPs than 28nm Planar and yet the Kaveri is still faster here than Ivy.

 

[SlowSpyder]

Using your Cinebench results, I decided to compare against my own system, as listed below in my signature. I used a Kil-A-Watt and measured from the wall for my power numbers, and used CPU-Z to obtain my voltages.

Total system power
FX @ 4.4ghz, 1.2250v
Idle: 120w
Cinebench: 242w
Score: 688

Ivy Bridge i5 @ 3.6ghz, 1.016v (stock speed, undervolted)
Idle: 37.4w (+/- 0.1w)
Cinebench: 69.0w (+/- 0.3w)
Score: 505

~

When comparing total system power consumption, my i5 comes out having approximately 2.6x the performance per watt.

It's not just the FX platform's high parasitic power draw skewing the results though. When you subtract idle watts from load watts (effectively isolating the CPU's power draw), and compare:

i5 Delta Watts: 31.6w
FX Delta Watts: 122w

The 3 generation old i5, in this light, has approximately 2.8x the performance per watt. Cranking it up to 4.0ghz reduces performance per watt slightly, but it still delivers 2.4x the performance per watt.

I highly doubt it's directly correlative, but AtenRa's testing with i3's shows Haswell as having ~33% better performance per watt than Ivy Bridge, and Skylake as having ~60% better performance per watt than Ivy Bridge. (watt hours to complete Cinebench R15 x264 benchmark). Also, I'm under the impression that HT CPUs have slightly better performance per watt than non-HT CPUs, which would make the FX look even worse when compared with an i7 (or i3, though they're not in the same ballpark in terms of performance).

Now, admittedly, your FX system still scores 36% higher in the benchmark, which has value in and of itself, but the power numbers don't look good at all.

FWIW

My FX was clocked at 4.4GHz, pretty high on the wrong end of the efficiency curve. I'm sure if I clocked it at something like 3.6Ghz I could get a huge jump in efficiency, and probably about match your score. I don't know I could be as efficient, but I know it'd be a lot closer than at 4.4GHz.

Look at the results, look at the i5's in the all core chart. Looks like a 4690k around 4.5-4.6GHz scores the same as my undervolted FX. According to Anandtech their i5 took ~180 watts at 4.6GHz. Put my FX in a 970FX board (less power use chipset), and a system without a big pump and 4x200mm fans and I bet it'd be much closer overall. What I'm getting at is any CPU can be pushed out of it's 'good' efficiency range for the sake of higher overall performance. The fact that I was able to run my silicon at 10% more clock than an FX8350 with a hefty drop in voltage, I thought that was pretty good.

I'm running this FX8320E through it's paces right now... I can tell you, it's a different beast than the FX9370, the FX8320E is having none of that 5GHz+ nonsense I ran my FX9370 daily with for a year. Heh. But maybe I'll see what it can do at lower clocks, how far I can undervolt it, my Kill-a-Watt is somewhere...

 

[Yuriman]

Yes of course, but I'm fairly confident that as you approach the same performance per watt as even Intel's 3 generation old chips, you're going to be well behind in performance, and there have been significant improvements in that area with each successive generation. Without further exploration we can't know for sure, but I'd really love to have a ton of AMD and Intel CPUs to make graphs and charts for. ^^

EDIT: It's interesting to me that the i7 in Anandtech's chart was drawing around the same power as the i5 at the same clockspeed, despite having a slightly higher voltage, and delivered ~11% higher performance.

 

[Madpacket]

My FX was clocked at 4.4GHz, pretty high on the wrong end of the efficiency curve. I'm sure if I clocked it at something like 3.6Ghz I could get a huge jump in efficiency, and probably about match your score. I don't know I could be as efficient, but I know it'd be a lot closer than at 4.4GHz.

Look at the results, look at the i5's in the all core chart. Looks like a 4690k around 4.5-4.6GHz scores the same as my undervolted FX. According to Anandtech their i5 took ~180 watts at 4.6GHz. Put my FX in a 970FX board (less power use chipset), and a system without a big pump and 4x200mm fans and I bet it'd be much closer overall. What I'm getting at is any CPU can be pushed out of it's 'good' efficiency range for the sake of higher overall performance. The fact that I was able to run my silicon at 10% more clock than an FX8350 with a hefty drop in voltage, I thought that was pretty good.

I'm running this FX8320E through it's paces right now... I can tell you, it's a different beast than the FX9370, the FX8320E is having none of that 5GHz+ nonsense I ran my FX9370 daily with for a year. Heh. But maybe I'll see what it can do at lower clocks, how far I can undervolt it, my Kill-a-Watt is somewhere...

My experience has shown 8320E's don't scale well with high voltage like the 8350/70's or 93XX series FX chips. They seem to top out around 1.3 to 1.35 volts. They're more about power efficiency with a respectable but smaller overclock (4.2 - 4.5Ghz) The nice thing about the 8320E's is how well they undervolt at stock speeds though. They act more like 2600K series Intel chips when undervolting where they become very energy efficient.

I'm not sure exactly what AMD did to the E series chips but I agree they behave quite differently.

 

[ShintaiDK]

Compare the performance and perf/watt of 22nm FF Core i3 3225 vs 28nm planar 55W A8-7600

Did you measure the actual CPU usage? Because it somehow dont seem to align with how review sites ended up. Excuse for no 3225 there, but the world had moved on. However I doubt 3225 to 4330 made that big of an impact.

http://techreport.com/review/25908/amd-a8-7600-kaveri-processor-reviewed/12

The AMD A8-7600 is not only faster, it also has higher perf/watt with a node disadvantage. This is simple amazing, if Intel would do that everybody would talk about the second coming of Jesus.

That's not what your cherry picked homebrew charts show.

And how big a die is an IB i3 again? Its not even half the size of your Kaveri

I assume you used 2400Mhz with the Kaveri and 1333 or 1600Mhz with the IB? And remember the IB still got ~40% faster ST. Same reason you can play DX11/DX12 games on the i3, but struggle on Kaveri.

You really thought nobody would notice the same "benchmark case" that's been reused ever since the FX release to try make AMD look better?