First impression before next week - FX mania

[csimon]

http://www.techpowerup.com/191895/am...0-pricing.html

Sweet!

If these can do 5ghz on AIO water cooling I'm sold

Test it!

Yes, now it's becoming attractive for those already setup for it.

[FlanK3r]

Im fighting with stability test now. Seems good, but scaling after some voltage/clock barrier is not much higher...But definitely not bad after all . Better than my FX-8350 in rig a bit

[Johan45]

Im fighting with stability test now. Seems good, but scaling after some voltage/clock barrier is not much higher...But definitely not bad after all . Better than my FX-8350 in rig a bit

I just posted this in your other thread but I'll give a bit more detail. I had to loosen the read to read write to write and read to write on the ram. this was almost like flipping a switch for stability on my 9370

[FlanK3r]

Storm chip is coming...

What the hell is this chip?!...

[Johan45]

WOW is all I can say!! Sweet silicon !

[FlanK3r]

can you try, if your FX can run Cinebench at similar voltage?

[Johan45]

Sure just give me a few minutes. Been playing around this morning. Are bus speed important?

[FlanK3r]

not much

[Johan45]

Ok here it is I made it to 5077, 5100 crapped out on me

[FlanK3r]

hmmm, very nice, the same ambients, around 22 C? Looks as next 8+ GHz chip

[Johan45]

hmmm, very nice, the same ambients, around 22 C? Looks as next 8+ GHz chip

It's a bit cooler here today maybe 18c. As for the 8 Ghz stuff i'll leave that up to you flanK3r. I'm not setup for that and still haven't decided if I want to venture down that road!! I'm always looking to make it go fast but..... Not sure if I can manage all that Mad Scientist stuff. SSomeday I might venture into it but not in the near future. I don't even know where I could find the LN2 here in Canada and then there's the transportation laws. Classed as dangerous goods here. This would take some research on my part.

[FlanK3r]

You right. In Canada people saying "its close", but in Czech Republic is "not far" around few meters or max few kilometers. In Canada I think it is 100km

[Johan45]

Gee 100KM eh? That would make London within distance. They should have it somewhere there. Ya Canada is a pretty big place. I can drive all day and still be in Ontario.

[FlanK3r]

Finally is the day of FX mania.



HWInfo and information about temps, voltage of first tested chip - this is FX-8300


At stock is FX-8300 in performance near the older FX-8150. Vishera is better in IPC and specially in games.

Cinebench R10 and R11.5



For all benchmarks I used decent OC at 4762-4763 MHz, but this clocks are not full stable I mean. So some average CPU....
4762 MHz benchable OC, Sisoft Sandra










Wprime 1024M with win7 64-bit with 2CU/4C software (not hadware shutdown)

wprime 1024M with 4CU/4C. With software assignment of process


Old one 3D Mark2005, GPU decent OC only


AIDA 64 memory bandwith (write, read, copy) at 4763 MHz and 2162 MHz DDR3 RipjawsX




AIDA CPU Queen


AIDA 64 Photo


AIDA 64 CPU AES


AMD Overdrive benchmark


Cinebench R10


hmmm, new Cinebench R15 at 4763MHz


HWBOT PRIME benchmark


MaxxPI 32M, scaling much better than old x87 superpi


Multicore Pi benchmark 4763 MHz


Nuclearus benchmark


One part of ScienceMark


Do you know ROG benchmark? He is long, stressed one part very well and including work with photo, video compression and multitasking



The maximum
Max Cinebench was able aroud 4815 MHz only :-(. This chip was very hot, but working OK up to 80 C in Coretemp!


Superpi 32M at 4965 MHz


And now something about quality of CUs. I tried it with SUperpi 32M and later with validation.
CU1 the best one


CU2 average


CU3 average


CU4 the worst


Superpi 1M max was around 5117 MHz and I want not used more than 1.55V for this chips


Validation 5317 Mhz with screenshot

[FlanK3r]

FX-8320 is second fighter from oldschool batch 1229PGN. I have two 8320 but this one I do not tested before. I can say this one is worse than first one...LLC I set as high, so real voltage was near the CPU-Z info.



ROG realbenchmark at 4766 MHz as warmup


AIDA memory and cache around 4900 MHz -warmup 2


X264 FHD benchmark at 4891 MHz and 4930 MHz



Most of benchmarks I was run at 5037 MHz and CPUNB over 2500 MHz (1.25V)

3D Mark06 at 5037 MHz


3D Mark11 at 5037 MHz


Cinebench R10


HWBot PRIME


MaxxPi 32M


benchmark Nuclearus


I love Cinebenchs, R11.5


First time we can see results of Truecrypt benchmark, stressed at 70 C in Coretemp!


Winrar benchmark


Wprime 32M and 1024M



Max benchmarks
Superpi maxed around 5162 MHz (32M, 1M I tested not for max)



Max validation after was around 5380 MHz

[FlanK3r]

So, what do you mean? New FX-9370 and old motherboard? It be working? Yes!

Im use for benchtable old and good one Asus Crosshair V Formula. This is really the solid board I think Asus make the best boards for AM3+. Cooling in all setups was AMD Allinone Cooler (Antec 920). Ambients between 20-22 C.



My FX-9370 is 1311PGN batch. This chip impressed me at begining but not all is so clear...
I began with Cinebench R11.5 at 4940 MHz. What this chip can do?

1.44V no problem...


1.42V? OK!


1.355V!! 4940 MHz


Cinebench R15 is a bit harder for CPU, I must jumped a bit with voltage


X264FHD benchmark with 4940 MHz and 2600+ CPUNB.


Impossible...CU4 is very bad and higher volts helps not much :-(. This CU4+temps kills all potential of this CPU..After few minutes of test I got hardware error.


With higher CPUNB clocks is CPU more sensitivity for volts, need more juice, limit is near 2800 MHz with 1.35V CPUNB


Usually, CPUNB 100 MHz lower help in CPU clocks. Cinebench at 5066MHz/2635MHz CPUNB


With this frequency I tried more benchmarks:

3D Fire Strike


3D Cloud Gate


3D Mark11


Fritzchess benchmark, easy over 16 000 points


Cinebench R11.5


Wprime 32M and 1024M


The maximum

Cinebench R11.5 at 5143 MHz


Superpi 32M test1 at 5262 MHz


Superpi 32M test2 at 5397 MHz


Superpi 1M at 5470 MHz!


Max validation, the best was CU1 (core 0+1) and worst last one CU4. CU3 was similar as CU1. 5647 MHz


hmmm, unvalid try. 5675 MHz (as always 1.555V as my maximum)

[FlanK3r]

And there is it, last one! Flagship of AMD, 9590!


5 GHz benchmarks
3D Mark11


AIDA64 memory and cache benchmark


AMD Overdrive bench


Cinebench R10


Truecrypt


Winrar bench


x264FHD benchmark



Overclocking benchmarks
5142 MHz 3D Mark Fire Strike


5142MHz 3D Mark11


5142 MHz ROG real benchmark


5242 MHz x264FHD


5242MHz AMD Overdrive benchmark


5242 MHz 3D Mark05


5242 MHz 3D Mark Fire Strike


HT link vs Cinebench score 26xxMHz/28xx MHz/30xx MHz




We want more! We want AMD enthusiast, ahuuuu!

5318 MHz Cinebench R15


5318MHz Fritzchess benchmark


5322 MHz 3D Mark06


5322 MHz MaxxPi 32M


5344 MHz Wprime 1.535V


5344 MHz Cinebench R15 1.544V 858 points


5344 MHz Cinebench R11.5 with 9.16 points


holly FlanK3r! 5371 MHz 1.55V Cinebench R15 run. This is monster chip, I need better cooling!


5344MHz Cinebench R11.5 second run 9.25!


The AMD "Spartans" never give up! My R11.5 "air" AMD record 5371 MHz 1.55V and 9.28 points!


5374 MHz HWbot PRIME benchmark


The next maximus...OK, not in 32M

Easy, easy with superpi 32M at 5418 MHz


5445 MHz Superpi 32M, but time is not much good (it was last benchmarks before borrow it back)


5500 MHz Superpi 1M


5554 MHz superpi 1M 1.555V


What about 5580 MHz Superpi 1M 1.555V "air"?
?

This CPU is madness 5747 MHz


5776 MHz CU2


All Core Units can idle at 5747 MHz, this is very good.

The best is CU2 (core 2+3)
5833 MHz!!

http://valid.canardpc.com/j2rf2f






Hmmm, what missing? Test stability? Ok, there is it:

Very nice 1.476V and 5033 MHz


1.535V for little jump up to 5092 MHz. More is impossible beause to hot for stability. But I found, every from all tested chips was hot else. Most of the chips has the best CU2 and all chips has worst CU4.

[csimon]

FlanK I'm at a loss for words. Awesome performance. What processor is the 4740 exactly?

[i found nemo]

5.5ghz air?! oh man that's crazy.

[FlanK3r]

Csimon: this is FX-9590

Yes, i was suprised from 9590, with better cooling it could be better! But I was shocked from 5090 MHz stability (30minutes). Never have before similar chip AMD/Intel

[Johan45]

Nice work FlanK3r. That 9590 really is a show piece isn't it? Wow that's a ton of work you've done! Congrats
I've still been kicking around the 9370 to see what it's limits are. So far with my setup I can clock higher in benches with my 8350, it'll take a ton of volts to do it but the 9370 locks up when I put the volts into so I think I'm reaching my thermal limit sooner with it.

[felix_w]

@Flank3r

I thought you had a custom water loop , didn't you ? Or it was rental ?

If you have, try it asap, don't let this chip wait!!!!!!!!!!

[FlanK3r]

no, I have not installed custom water loop. I have one older in the closet. But Im lazy building this PC . Maybe with Steamroller FX. At benchtable, thats mean every time another CPU/MB Im use simply Gelid Tranquillo or Noctua NHD14 or AMD allinone liquid.

[mcnbns]

I guess this thread almost puts to rest the theory that all FX 8300s clock better than 8350s.

Maybe it's not like the 2500+ XP-M vs regular 2500+ of old because the default clocks on the 8300 are so much lower than the 8350's.

[Johan45]

I have done a bit of a comparison myself between my 8350 and my 9370. The biggest difference I found was the voltage needed for similar clocks. I have a better cooling solution than FlanK3r at the moment so I pushed the voltages a bit higher. This is only a chart no fancy screen shots or anything like that. It can all be verified if you check my HWBot profile. I'm not making any of this up.
My conclusion is that the 9370 is a tigher chip and heats up really quickly as compared to the 8350. All in all performance is very similar and only restricted by the cooling. Also I have had this 8350 for months and did the majority of my learning FX on that chip. So there may be room for improvement on the 9370 with some more time but not much.