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

[undone]

The new FX-series microprocessors from AMD, which are due to be formally introduced on September 1, 2014, are the FX-8370 and the FX-8370E. Both chips feature eight cores, 4.10GHz/4.30GHz clock-rates, 8MB L2 cache, 8MB L3 cache and dual-channel DDR3 memory controllers. The FX-8370 will have thermal design power of 125W, whereas the FX-8370E will feature TDP of 95W. Both microprocessors will be priced at $189 in mass quantities.

In addition, AMD will also slash prices on existing FX-8300-series chips and introduce the model FX-8320E with 95W TDP.

http://www.xbitlabs.com/news/cpu/dis...rocessors.html

[SNiiPE_DoGG]

Really? literally 2 days after I buy a 8350 lol.... I assume there would be no benefit to returning my cpu and waiting for this if I'm OCing except for the 30 extra bucks in my pocket?

[vario]

Really? literally 2 days after I buy a 8350 lol.... I assume there would be no benefit to returning my cpu and waiting for this if I'm OCing except for the 30 extra bucks in my pocket?

Well the TDP seems weird, 4.1 base and 4.3 turbo and 125 and even 95W is like impossible on visheras.Unless theyre like BEST OF THE BEST or theyre just lying ,so they will be throttling as hell on default and high load.Mine at 4.0 does like 160 or 180W.

[FlanK3r]

you have wrong settinggs or something...125W with 8350 is not the problem. Its possible get with undervolting and stilll stable setings around 100-110W TDP of FX-8350.

[vario]

you have wrong settinggs or something...125W with 8350 is not the problem. Its possible get with undervolting and stilll stable setings around 100-110W TDP of FX-8350.

firstly, this thing goes faster than 8350, secondly at 95W.
And lets be clear here flanker, you time and time again talk about "stability" but when i ask you does that mean ANY application for ANY amount of time, you say NO.
And one thing more, these things like to cheat with throttling.So in order to make the frequency stable i have to turn off APM and boost up LLC ,and even though my voltage is below stock (1.332v) LINX or prime consumption is HIGH.When i turn ON APM and set default to other settings, MAYBE the thing will be around 130-140W BUT its gonna throttle from time to time (and also be more prone to random freezes)

[FlanK3r]

firstly, this thing goes faster than 8350, secondly at 95W.
And lets be clear here flanker, you time and time again talk about "stability" but when i ask you does that mean ANY application for ANY amount of time, you say NO.
And one thing more, these things like to cheat with throttling.So in order to make the frequency stable i have to turn off APM and boost up LLC ,and even though my voltage is below stock (1.332v) LINX or prime consumption is HIGH.When i turn ON APM and set default to other settings, MAYBE the thing will be around 130-140W BUT its gonna throttle from time to time (and also be more prone to random freezes)

The "NO" is most of time with busy style. Because Im tetsing with one benchtable some boards and another CPU time the time. But in last months I tested power consumption and 30minutes stability test. The thread about it is here:


http://www.xtremesystems.org/forums/...hlight=FX-9590

[vario]

The "NO" is most of time with busy style. Because Im tetsing with one benchtable some boards and another CPU time the time. But in last months I tested power consumption and 30minutes stability test. The thread about it is here:


http://www.xtremesystems.org/forums/...hlight=FX-9590

Flanker i know we differ in this opinion, but 30min AOD test is barely semi stable setting :-/ .As i do often some heavy stuff and multitasking i have to be certain that the thing will be stable for for example 12h jobs ,and theres a difference between something that behaves at first glance as stable and something that really is.Anyhow i just checked, right now at 4.0ghz 1.344v load theres a 175W difference between idle and load and it slowly climbes up .However , i gather you have 9590 at your disposal,should be great silicon, then do a test if youre gonna be able to go down to 95W at 4.1 continuous load of prime 95 x64 .I gather, you wont.

[NEOAethyr]

It's been such a long time but I've seen some cpu's cheat on prime before, not maxing out to the set multiplier.

So far from all the progs I've tested with, linx/intelburntest works 2nd best.
A half hour or more of watch dogs gameplay is even more stressful then over 12hrs of those 2 progs maxed though.
I have not gotten around to see if prime beats then in stress testing though.

On amd, s&m's memory tests, the full set of 4 of them, on normal/avg length did quite well for checking the cpu but it was never the end all final say in it.

If they can get under 180w's, 5ghz, maybe, maybe..., they can get the mid range budget buyers back.

I think they underestimate the enthusiast, it's us that actually recommend things to the avg user...
If we don't like it, then we don't tell anybody to buy them.
I THINK it's the only reason they sold any reasonable amount in the 1st place.

I was reading some threads on guru3d about the new cpu batches.
Alot of peeps complaining about amd's amount of core's vs single core perf.
Bugs the heck out of me, 8 cores vs 4 cores, there's no diff, it's just marketing.
I wish we had more cores...
And yet I know we need more perf, and we definitely need lower power.

It really has turned from an honest debate to a good laugh.
I think it's funny as heck sometimes, yet slightly depressing.

Seen a quote, in it stating the state of the art tech pcie2.0 lol.

It's all a good laugh, and sometimes it's interesting to see if they have improved the batches just a wee bit.
But in the end, you can pretty much guess the end results.
I think we gotta another year or so to wait 'till interesting things maybe.

Ddr4 is gonna be crap for a while anyways.
3200mhz at 16t timings, which apparently can't be done with 4x8gig sticks let alone 8x of them.
Amd I think will do alright as long as they can hold on and not go bankrupt.

[Greg83]

what is the power efficency of the device and the age of the device and the margin of error of the device you're measuring the power consumption of the processor with?
to say the TDP measure by million dollar devices is incorrect.

what is the power efficency of a consumer device PWM and mosfets. capacitors, they are so small nothing like even 300w power supplies.

what is age of your devices at time of measurements?

and as far as i always knew . thermal dissipation properties. were literally speaking about the heat of which is being released and must be cooled by a heat sink.

such as how a CFL bulb produces power on its own and holds power. theirfor it takes longer to disipate the power. then its actual hydro usages. opposite can be true too.

the heat to cool the TDP . can be less then its power consumption. cause it used some of that energy. it was lost and some was reused and grounded out and sent back through the negative channels.

[NEOAethyr]

you have wrong settinggs or something...125W with 8350 is not the problem. Its possible get with undervolting and stilll stable setings around 100-110W TDP of FX-8350.

I can't speak for the 8xxx and 9xxx's but the 1090t I had could run something like -0.625v(I can't remember if this is right...) at 3.25ghz and +0.1v from that (turbo) at 4ghz.

Eh crap I don't have it writte down.
Might of been -0.1625 or -0.0625, it was whatever the lowest offset was on the ch4f.

Going much lower, even as far as to redo all the multi's too but...
Anyways it had probs with high llc at lower voltages (ie lower power state).
Low lcc it was fine at extreme low voltage (vid's anyways).
Set with that msr tweaker prog whatever it was called.
But I never ended up using that prog for 24/7, with all the work I put into it it was still a ways to go so I gave it up.
Didn't have a separate cpu vid for each core on each power state either so whatever anyways I guess.


It was a while go but someone here posted a bit of info on the power states.
I just remember posting in the thread, that's it lol.

From that info I gather they screwed themselves on power management from the get go on the 8xxx and above.
Maybe all they did with these new batches was introduce new bios code and a new cpuid for the lower power states :\.
They could of of least gave us a new chipset, that used to be the norm.
Well sometimes they would recycle them and relabel, but they've had the 890fx for so long now.

[vario]

I can't speak for the 8xxx and 9xxx's but the 1090t I had could run something like -0.625v(I can't remember if this is right...) at 3.25ghz and +0.1v from that (turbo) at 4ghz.

Eh crap I don't have it writte down.
Might of been -0.1625 or -0.0625, it was whatever the lowest offset was on the ch4f.

Going much lower, even as far as to redo all the multi's too but...
Anyways it had probs with high llc at lower voltages (ie lower power state).
Low lcc it was fine at extreme low voltage (vid's anyways).
Set with that msr tweaker prog whatever it was called.
But I never ended up using that prog for 24/7, with all the work I put into it it was still a ways to go so I gave it up.
Didn't have a separate cpu vid for each core on each power state either so whatever anyways I guess.


It was a while go but someone here posted a bit of info on the power states.
I just remember posting in the thread, that's it lol.

From that info I gather they screwed themselves on power management from the get go on the 8xxx and above.
Maybe all they did with these new batches was introduce new bios code and a new cpuid for the lower power states :\.
They could of of least gave us a new chipset, that used to be the norm.
Well sometimes they would recycle them and relabel, but they've had the 890fx for so long now.

Well yea, X6`s were great underclockers to a point.I had 1055T ,2.8Ghz stock at 1.4v ,however it worked 24/7 at 3.5ghz and 1.25v ,great chip.Going for the 4ghz required massive increase in power tho.Not really seeing this trend in FX tho, i can get 4ghz at a bump lower voltage, but power consumption is still high and if i want a non throttling one, it exceeds nominal by a big margin.
APM ON means cores fluctuating between 3.2 and 4.0, APM OFF and HPC ON means rock stable and 4.0 but power consumption is around 50W more.Some people doesnt seem to know that with FX your overclock may be only approximation of teh frequency you get.Anyhow.Will see. Maybe AMD did a new revision? But doubtful.

[cab]

So what is the difference between FX-8370 and FX-8370E? Do they perform the same? Are they both black edition?

[FlanK3r]

yeah, both are with unlocked multipliers. The diference is the base clock (without turbo). You will see very soon ...

[MattiasNYC]

yeah, both are with unlocked multipliers. The diference is the base clock (without turbo). You will see very soon ...

I thought the difference was the opposite; using less power at the same base clock but some difference with turbo...

[cab]

yeah, both are with unlocked multipliers. The diference is the base clock (without turbo). You will see very soon ...

lol, you have too many CPUs already.

Same performance with lower tdp seems impossible, unless AMD has perfected the process and can reliably undervolt the FX-8370E ?

[ilkkahy]

I wonder could amd make FX9590E with 5/4ghz clockspeeds staying within 125w TDP? Does anybody know how much FX9590 takes watts when running single thread at 5ghz?

[BeepBeep2]

Performance is in line with the base frequency (3.3 GHz) for the 8370E...a little slower than FX-8150 in multithreaded tests.

[The Stilt]

I wonder could amd make FX9590E with 5/4ghz clockspeeds staying within 125w TDP? Does anybody know how much FX9590 takes watts when running single thread at 5ghz?

You mean the base clock would be at 4000MHz and the boost up to 5GHz?
The trouble in clocking the cores that high is the fact that the design reaches it's critical point at 4.6GHz region.
Beyond that the efficiency drops significantly as the voltage needs to be increased heavily.

I?d say it would require atleast 140W TDP even on the newer revision, but I can see if I can take some actual measurements of that kind of scenario.