AMD Readies FX-8370, FX-8370E Microprocessors.

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
Yes and no. It can happen that way, but only with a very mature process node which has had its functional and parametric yields optimized to a tight(er) distribution. Which in general is where all process nodes end up once they have been in the hands of the fab engineers for 6-8 quarters, if not sooner.

Okay, hold on - your original statement was that the chips will use similar power consumption when operated at the same frequency, voltage, and temperature. So the coefficients in the dynamic and static power consumption equations will be close to the same.

But SOFTengCOMPelec was making a much stronger statement, that the chips will have the same frequency/voltage curves, max frequency and voltage and so on. When is this ever really true even with a mature process? If this were the case Haswell, for example, they would all overclock the same under the same parameters, when in practice there's a pretty huge amount of variation.

Same thing with mobile SoCs, if you look in kernel source you can see tons of different bins describing separate voltage/frequency curves.
 

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
As bolded above, the cpus are FACTICIOUS, and don't exist.

It was just an example, so that I could ask questions about wafers/cpu quality vs prices, and other stuff. I needed to keep the $49 at 8 cores (in practice it would probably be 2 cores), so that the clock frequencies and TDP were for comparable parts.

Awww... I didnt see that,I was just so excited for the $49 price I didnt even bother to see what else was written.I was completely jumping for joy for a 8 core cpu for that price point. :( Oh well back to the drawing board..
 

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
Awww... I didnt see that,I was just so excited for the $49 price I didnt even bother to see what else was written.I was completely jumping for joy for a 8 core cpu for that price point. :( Oh well back to the drawing board..

No problem, an easy mistake to make.

Earlier in this thread, the 8320E (8 Cores) for $139, is not a bad price for an 8 core device. E.g. Running lots of VMs at the same time.

By the time mainstream cpus are $49 for 8 cores, the $299 (top non-extreme consumer parts) will probably be something silly, like 32 or 64 cores.
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,859
4,835
136
If this were the case Haswell, for example, they would all overclock the same under the same parameters, when in practice there's a pretty huge amount of variation.

That was the exemple i had in mind since there s 10% dispersion at least between the ocking results, and these are chips that did pass the quality tests, how much were out of specs and rejected.?..

At thoses node levels caracteristics dispersion is a big issue because it reduce yields that are already significantly lower than with previous processes for purely lithographic defects that render the chip totaly unfunctionnal.

Indeed when a foundry say that it has 90% yields it doesnt mean that 90% of the waffers surfaces yields functional chips, this means that it has reached 90% of its targeted yields at wich point the marginal improvement become cost innefficient.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Okay, hold on - your original statement was that the chips will use similar power consumption when operated at the same frequency, voltage, and temperature. So the coefficients in the dynamic and static power consumption equations will be close to the same.

But SOFTengCOMPelec was making a much stronger statement, that the chips will have the same frequency/voltage curves, max frequency and voltage and so on. When is this ever really true even with a mature process? If this were the case Haswell, for example, they would all overclock the same under the same parameters, when in practice there's a pretty huge amount of variation.

Same thing with mobile SoCs, if you look in kernel source you can see tons of different bins describing separate voltage/frequency curves.

There is a difference between saying two chips will each have essentially the same Fmax at the same Temp, Vcc and Pmax, versus saying that (within reasonable limits) two chips will exhibit nearly identical Power footprints when operating at the same GHz, Temp and Vcc.

It is the difference between talking about the edge of the shmoo plot (the line between pass and fail) versus discussing points that fall within the pass area itself.

If I haven't been clear in this distinction then I apologize.
 

Exophase

Diamond Member
Apr 19, 2012
4,439
9
81
There is a difference between saying two chips will each have essentially the same Fmax at the same Temp, Vcc and Pmax, versus saying that (within reasonable limits) two chips will exhibit nearly identical Power footprints when operating at the same GHz, Temp and Vcc.

It is the difference between talking about the edge of the shmoo plot (the line between pass and fail) versus discussing points that fall within the pass area itself.

If I haven't been clear in this distinction then I apologize.

On the contrary, I think this distinction was clear in your first post, and this post basically mirrors the post you were responding to. Where the confusion comes from is where SOFTengCOMPelec responded with this:

If thoroughly tested, nearly all the above devices in practice would achieve approximately 3.9 GHz at 95 W (TDP) measured power consumption under high load on all 8 cores, with maybe only a 5% variation, due to the qualities of modern chip manufacturing ?
To which you more or less said yes, if the process is mature enough. But for those chips to all have (nearly) the same measured power consumption (as opposed to just being labeled with the same TDP) they'd have to be running at about the same voltage (and same temperature, but we can give it that). Maybe if you overrode the dynamic voltage scaling this would be the case, but at the very least there'd be a range of minimum voltages that different CPUs tolerate.

To make it clearer, he followed that up by stating that it'd mean that all processor segmentation was artificial and binning was no longer a thing.
 
Last edited:

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
It is capped at about 70-75W, it doesnt consume as much as its rated TDP, isnt it, but of course we know that Intel CPUs do not always consume their rated TDP when pushed, do you know why ?..

Because Intel choose to throttle the GPU instead of the CPU while AMD choice is to never throttle the GPU and rather cut the CPU frequency, Kaveri CPU throttling is only when both GPU and CPU are fully loaded.

That is, you would had known it had you read said hardware.fr article.

http://www.hardware.fr/articles/915-4/turbo-ctdp.html

Wow, you made this an Intel thread huh?

Well it's not, and anything you post trying to deflect doesn't change the fact that the article you linked shows the AMD CPU dropping below base clocks when operated within specifications.

Get it?
 

Abwx

Lifer
Apr 2, 2011
11,859
4,835
136
Wow, you made this an Intel thread huh?

Well it's not, and anything you post trying to deflect doesn't change the fact that the article you linked shows the AMD CPU dropping below base clocks when operated within specifications.

Get it?

Not at all, i explained technical choices, when CPU and GPU are pushed you have to distribute the power budget and there s different way to do it, you can trotthle one of the two components of even both, in that case AMD choice was to throttle the CPU and not the GPU, other may have a different approach and are not branded suspect because they do so, i did use intel as exemple of a different implementation, it could had been Qualcomm if i knew how they manage their products thermal management, so actualy it s you that are making this thread an intel/AMD debate because you failed to perceive the technical aspect and relied instead on a supposed deflection that exist only in your mind, indeed you should had stayed on the technical debate...
 

Gikaseixas

Platinum Member
Jul 1, 2004
2,836
218
106
Games - 80/20 in favor of 2500K
Single threaded apps - 2500K
Multi threaded apps - FX 8370E
Power Consumption - if you really care then 2500K

Conclusion: Don't do it unless you have a board laying around. You would be better served by getting a 4790K
 
Last edited:

Enigmoid

Platinum Member
Sep 27, 2012
2,907
31
91
Interesting. Anandtech's article/source mentioned only boost clocks, not base clocks.
 

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,002
126
I have a spare 990FX AM3+ board. And 8320E or 8370E sounds pretty tempting. I was leaning towards an FX6300 for it, and still might depending on where prices end up after these new CPU's are available. But, an 8370E with no overclocking and a quiet cooler would probably be enough for a few years yet for my eight year old.
 

JDG1980

Golden Member
Jul 18, 2013
1,663
570
136
Interesting. Anandtech's article/source mentioned only boost clocks, not base clocks.

They've done this on video cards for the last couple of releases. Must be a marketdroid push. IMO, if they're not going to announce actual base clock rates, Anandtech should put "0" in the chart. For CPUs, this missing information is absolutely unacceptable.
 

crashtech

Lifer
Jan 4, 2013
10,693
2,290
146
Pfft, when I was an 8 year old we had to make do with a Pentium II! ;)
When I was eight...

640px-SR-50_early_TI_calculator.agr.jpg
 

Bradtech519

Senior member
Jul 6, 2010
520
47
91
The FX-8370E would be a decent upgrade over my FX 8350 considering I crunch at 100% load 24/7 most of the time. Selling the FX 8350, and purchasing a 8370E could save some on the electric bill.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
The FX-8370E would be a decent upgrade over my FX 8350 considering I crunch at 100% load 24/7 most of the time. Selling the FX 8350, and purchasing a 8370E could save some on the electric bill.

Assuming it can keep the same clocks while you crunch. That however is to be seen.

Its already stated that it will turbo for shorter periods.
 
Aug 11, 2008
10,451
642
126
Assuming it can keep the same clocks while you crunch. That however is to be seen.

Its already stated that it will turbo for shorter periods.

Yea, I agree. I would wait for some direct power consumption and performance tests before assuming the TDP numbers translate directly to power savings. The whole line-up just seems weird if they in fact have two cpus at the same price, same clockspeed, and same TDP.
 

Bradtech519

Senior member
Jul 6, 2010
520
47
91
Assuming it can keep the same clocks while you crunch. That however is to be seen.

Its already stated that it will turbo for shorter periods.

I hate to keep putting money into AM3+. I married myself to AMD with a heavily discounted price for my FX 8350 setup. I'll put it off to wait and see some reviews for sure.