AMD Zambezi news, info, fans !

[pumero]

Well, it's actually made by Asetek anyway, who make most of the sealed liquid cooling solutions (along with CoolIT), the Kuhler 920 seems to outperform most of the air cooling solutions for the same money, so it should be a good solution for a decent OC on these chips!

If you are deaf it's surely a good solution - otherwise, think again.

I did some tests with those water cooling solutions and yes, their performance is not bad. However, the H70 and the Antec 920 for example come with two fans which run at high speeds, which ensures good airflow through the radiator, but create an (in my opinion) unacceptable noise level.
With the Antec 920 it was necessary to have both fans at ~2200rpm (max. is 2500rpm) to keep a 5GHz Sandy Bridge overclock at ~80°C or slightly above, while 2500rpm brings you barely below 80°C.

Sure, if you compare the raw performance at max settings of the Antec 920 with a NH-D14 or Silver Arrow at stock, it's clear the Antec solution will win. If you like to have a system that sound's like a turbine, go for the Antec but said air coolers give you a solution that might be slightly weaker at stock, but you can actually work next to them without the need for ear protection.
However, using similar powerful fans on those coolers I highly doubt that there will be much of a difference between them and the Antec 920/H70.

In my opinion, if you already have on of the best air coolers at hand, it's better to save the money and go for the FX cpu without the water cooling solution.

[Opteron146]

correction Prefetch, or is it trace cache. all i know after reading is it's 8Kbyes for the module and 4Kbytes to each core. -__-

i am so lost on bulldozers lol

You maybe mean the write (combining) buffer to the L2 which is needed because the L1 is write through?

[radaja]

does anyone have access(member with 10+ posts)to the website corescn.com?
they have a thread about the FX-4170 and i want to know if there are any benchmark results in it.
JF-AMD said if you care about single threaded performance
then the lower core higher clocked models would be a better comparison.

i was able to copy the little picture that pops up in the bottom right hand corner though


this is the chip i find most compelling,4.2GHz stock is so sweet

[dess]

@Leeghoofd: Could you please check if the all-cores-on turbo is able to activate itself during the multi-threaded benches and apps?

[Leeghoofd]

will do tomorrow, finalising ram tests now...

[Macadamia]

You need 10 posts for that Corescn site's CPU section, and I'm not in the mood to spam 10 posts in Chinese right now

[radaja]

You need 10 posts for that Corescn site's CPU section, and I'm not in the mood to spam 10 posts in Chinese right now

i just spent the last hour trying to register and finally got it done,but now it takes me forever to even make one post
having to switch between the original page and google translate to do the verifications and capcha's for a post is insane

[Macadamia]

Oh fine I'll do it.


Edit: The guy's still on 08XX BIOS, so far rather mediocre results on most benchmarks, but...


Cine11.5 doesn't look that bad, it performs ~ PhIIx4 925 -- 945 territory.

[haylui]

i just spent the last hour trying to register and finally got it done,but now it takes me forever to even make one post
having to switch between the original page and google translate to do the verifications and capcha's for a post is insane

i saw you there
you told the people that you're come form USA and with a difficult to understand job position?

[haylui]

yes...10 posts in corescn

[cesariuth]

Someone have problems with turbo core on FX-8150?

[haylui]

FX-4170 SuperPI


Fritz Chess benchmark


3DMark Vantage


AIDA 64


WPrime


7zip


Cinebench R11.5

[radaja]

i saw you there
you told the people that you're come form USA and with a difficult to understand job position?

well thats how it came out but thats not what i was trying to say
i think it should have been,"i forgot to add,i'm from the US so my posts my be difficult to understand"
but thats google translate for ya

[mzs_biteme]

Cinebench R11.5

WTF!!!!!! Why CB shows 4.22GHz FX4170 faster than 4.22GHz 8150??????

[EniGmA1987]

AIDA 64

Need to see the NB speed on that, cause that is TERRIBLE memory latency
Also, wasnt B2 the not good stepping?

[haylui]

well thats how it came out but thats not what i was trying to say
i think it should have been,"i forgot to add,i'm from the US so my posts my be difficult to understand"
but thats google translate for ya

replace "posts" with "replies" that could be better.
lol.

[haylui]

do you think you guys need the hardware setup for the FX-4170 benchmark?

Motherboard: Asus CrossFire V
Bios ver: 0811
Memory: Kingmax DDR3 2400 (lower to 1866)
Display card: MSI 5970
HDD: OCZSSD2-2VTXE120G
OS: Windows 7 64bit SP1

[Evantaur]

WTF!!!!!! Why CB shows 4.22GHz FX4170 faster than 4.22GHz 8150??????

either that or it's a fake

[itsmydamnation]

Need to see the NB speed on that, cause that is TERRIBLE memory latency
Also, wasnt B2 the not good stepping?

doesn't seem that bad, compared to my Nelaham

memory latency is 28% faster then tri channel 1333 7-7-7-20, this is actually really good, because high clock memory sacrifices latency for throughput and its dual vs tri channel.
L1 latency is the same
L2 is 43% slower but 800% bigger cache
L3 33% slower but is exclusive

we new long ago that L2 latency wasn't going to be awesome, but you can hold alot more data on a BD chip then a Nehalem/westmere/SB and avoiding needing to fetch from memory will have a far bigger impact then the difference between cache latencies as hopefully the really important stuff will be in L1.

edit: im running 21/22x180bclock and an OC'd uncore as well(cant remember what its running at)

[Dirk Diggler]

WTF!!!!!! Why CB shows 4.22GHz FX4170 faster than 4.22GHz 8150??????

Either its fake or the rest of the 8*** series were not as optimized as they shouldve been eg (latest bios etc)

[Baam]

Well new AGESA code BIOS was released today by most of the motherboard manufacturers. Might need to retest.

[tbone8ty]

this is getting so exciting now! Had re read last few pages twice to make sure I got everything!
watercooling!
sick ram multipliers!

my CH5 purchase is looking more and more like the right choice!

wondering if my gskill sniper kit will at least make 1000mhz now with bulldozer

my have to find a nice 2133 kit to match ;-)

[bamtan2]

FX-4170 SuperPI

does this superpi mean anything? my 5 year old core 2 duo with ddr2 does 1M in 25s.

[Manicdan]

super pi is crap, im going to the blue team!



waits to see how many people quote this thinking im serious...

[BeepBeep2]

Might be down to the fact the chip is tuned for raw GHz and the thermals are the limiting factor, thus cooling for everyone.

TDP is thermal design power, has nothing to do with power the chip pulls from the board but the heat it gives off. Therefore, a 125w chip should run about as hot as any other 125w chip assuming they are same architecture with same thermal sensor...(or you could simply use an external probe, like k-type bead or flat diode)

@SSJVegeta
It will beat the Silver Arrow and Archon.

The radiator is very similar if not the same as the Corsair H70. It will also have the same block design, chances are same fans. The Antec Kuhler is the same essentially to this unit, which has the rubber tubing instead of plastic.

Either its fake or the rest of the 8*** series were not as optimized as they shouldve been eg (latest bios etc)

1, it's not fake. That shows single thread performance. The 8 core most definately does not score less than 4 points, lmfao.
2, doesn't that CPU turbo to 4.3?
7% difference with ~3% clockspeed, I can only guess that the Win7 scheduler got luckier with the 4 core than with the 8, and that the more cores you add the slower your IPC gets. (Look at Thuban, yes I know it has less cache per core than Deneb but still)