The AMD FX-9590 "FX Centurion" Preview

[MacClipper]

After weeks of waiting, it finally arrived...

 

[haylui]

After weeks of waiting, it finally arrived...

after so much information has been revealed, NDA lifted but no more excitement...haha​

 

[MacClipper]

jus curious, does "CPU" circled in blue refer to temp at the motherboard cpu socket? if yes, then shudn't the cpu itself also be in that range too (actually slightly higher)...? eg. maybe abt 70-75C instead of 49.4C.

Dug up again, info specific to the ROG Crosshair series

And this is what it looks like for the FX-9590 after an hour of mainly web surfing and some Cinebench/WinRAR benching.

 

[MacClipper]

Advise for FX-9590/9370 (Centurian) users on C5F-Z/Sabertooth 990FX R2.0

 

[MacClipper]

Gist of above link for those too lazy to click...

AMD has validated the Crosshair V and Sabertooth 990FX as for use with its new FX 9590 and 9370 CPUs, however there is some advise we'd like to add for users upgrading (or even users running their FX-8350s at ~5GHz frequencies).

1) You MUST plug in the 8pin and 4pin CPU 12V power plugs: they are there for this kind of usage! A 5GHz FX CPU will certainly draw over the 300W limit on the 8pin cable. If your PSU has good OCT it will simply cut out and your system will suffer hard-reboots. If not, it will likely melt poorly constructed PSUs with cheap wiring under extensive stress, unless your PSU is specifically fitted with heavy-gauge (lower AUG rating) wiring.

2) Do not disable OCP if you're running 24/7, it will protect your CPU. If you're trying to run extreme voltage for benchmarking you will have to turn it off, but in that case we strongly recommend watercooling (or at least chilling) the VRMs.

3) As you'd expect, when pulling close to 400W the VRMs run hot. At a minimum active cooling from direct case airflow is therefore a must (and watch those temps). Still, we strongly recommend watercooling the VRMs (EK does a good block) and do NOT loop CPU into VRMs: create two loops. When you consider your loops take into account the heat you're trying to shift: we advise looking at >1 large radiators and/or even a chiller system is maybe worth considering if you're running several high-power items together.

HTH

 

[bhtan83]

but based on ur review the highest was 172W only.. unless its for the extreme OCing conditions.

 

[MacClipper]

You must be referrring to this line from above, read it again slowly and carefully...

1) You MUST plug in the 8pin and 4pin CPU 12V power plugs

Then check out the following large image of the "validated" Sabretooth R2.0 for a clearer picture on plugging BOTH 12V power plugs.

http://content.hwigroup.net/images/products/xl/160198/asus_sabertooth_990fx_r20.jpg


HTH

 

[MacClipper]

Load Temps
Max load DTS very similar to the Extreme9 and here ROG CPU (T0) temps seem very close to DTS too

 

[hlots123]

thx for the follow-up, mac
yah the temp readings with the asus board is easier to make sense of

side qn, any ideas why despite the low load temp, how come amd specs for max temps (when stated) are usually much lower than intel's? i thot silicon wafer tech shud be abt same for similar generation whether amd or intel wrt silicon/circuit degradation...

and the working range for the amd cpu package sensor (only valid above 40c?) seems so narrow.

 

[MacClipper]

Cannot compare like that cos quite different designs and platforms mah...

 

[MacClipper]

HWBot Prime
Bumped up scores from 6653 to ~6663 with lower latency 1866C8 RAM


Old score

New score

Nice!

 

[ragnarok95]

Nice! Getting better!

 

[MacClipper]

Running RAM at 2133C9 didn't achieve any higher scores. So yeah, I think that's a pretty decent score improvement.

btw, don't expect any single socketed Intel rig to come close to this score even for the enthusiast LGA2011 platform, much less the mainstream 1150/1155 rigs.

 

[Gattberserk]

Running RAM at 2133C9 didn't achieve any higher scores. So yeah, I think that's a pretty decent score improvement.

btw, don't expect any single socketed Intel rig to come close to this score even for the enthusiast LGA2011 platform, much less the mainstream 1150/1155 rigs.

So the FX5950 is faster than any Intel consumer CPU in this aspect?

 

[MacClipper]

That's what I had read.

Gattberserk, how about posting your 3770K @ 4.4GHz score to see if it is true? Thx.

 

[MacClipper]

Memory Bandwidth Scaling
FX-9590 on Crosshair V Formula-Z 1304


Corsair Vengeance Pro DDR3-2133, "Intel RAM"
- Samsung SS memory

Team Elite DDR3-1600
- PSC DS memory

HTH

 

[haylui]

MacClipper, any chance you disclose your detailed system spec for running the benchmarks?

 

[MacClipper]

hi haylui, almost exactly the same setup as earlier posted in #2 except for the components being tested.

OK, will just repeat it here


★ AMD FX-9590@5GHz | Crosshair V Formula-Z | NiCu HK 3.0-GTX360 | HIS HD7950 IceQ X²@stock | Enermax Platimax 1200W ★


RAM timings used in each test can be read from the AIDA64 screenshot itself.


HTH

 

[haylui]

hi haylui, almost exactly the same setup as earlier posted in #2 except for the components being tested.

OK, will just repeat it here


★ AMD FX-9590@5GHz | Crosshair V Formula-Z | NiCu HK 3.0-GTX360 | HIS HD7950 IceQ X²@stock | Enermax Platimax 1200W ★


RAM timings used in each test can be read from the AIDA64 screenshot itself.


HTH

thanks for the details system spec.

From AMD FX-9590 Review; Piledriver at 5GHz - Page 17

they shows 248W of total power consumption (measure from AC supply).
Could you run the similar test to see if their chip and your chip deviate?

 

[MacClipper]

If you take into consideration the many differences in the test configs, think you may be able to account for the ~70W gap

Difference...
AC voltages - 240V vs 120V
Mobos - Extreme9 vs Sabertooth
VGA - HD7950 vs GTX670
PSU - Platinum 1200W vs Gold 850W
Vcore - undervolted vs stock
APM - active vs ?
Windows Power Plan - Balanced vs Performance
Cooling - custom water vs air
Measurement - direct watt meter vs Tripp-Lite 1800W line conditioner
Load test - P95 Blend vs wPrime
etc...

Again, note that it's 248W for their whole rig off the wall vs the official 220W TDP just for the CPU alone.