AMD - Piledriver chitchat

[Biscuit]

Someone has got hold of an FX8350:

http://www.overclock.net/t/1318240/h...ak-vs-3770k-ht

They compared it when overclocked to 4.5GHZ with a Core i7 3770K at 4.6GHZ with HT switched off, however, remember the Core i7 CPUs have more L3 cache too than a Core i5.

Seems there is HT and non HT benches on there now. i7 smashes it with HT on!
Still an impressive indications if the performance is like that across the board, hope we don't have to wait much longer.

[watercooled]

i7 is significantly more expensive, the pricing of the top binned PD is clearly aimed at i5.

Anyway, if you look, the i7 isn't miles ahead apart from in the first benchmark which appears to be Winrar, but AFAIK it's not indicative of real-world compression speed.

Edit: While the PD chips performs OK in Cinebench, for anyone unaware it's important to remember it's compiled with the Intel compiler. It's debatable how much impact this may have but if you search for 'Intel Compiler Patcher', one of the first programs you'll see mentioned is Cinebench. I'm not sure if that 'patcher' works as it claims BTW, I'm just using it as an example. Unless I find proof otherwise, I don't consider results comparing Intel to AMD (or VIA for that matter) reliable.

[c.ruel]

If it equals an i5's performance and the cpu costs less alone NOT factoring in the motherboard then they will definately sell a fair few.

I wonder what the power consumption and overclocking is like!?

[scaryjim]

Hmmm, CAT's comments about processor optimisations in another thread just made me think of something - has any review site tried benchmarking Trinity under Win 7 and Win 8 to see if scheduler optimisations help performance? I know the early BD tests showed moderate improvements in lightly threaded loads between Win 7 and Win 8 dev preview - with the release preview already kicking around (and therefore - presumably - better optimised) I'm surprised no-one's tested this...

[CAT-THE-FIFTH]

Hmmm, CAT's comments about processor optimisations in another thread just made me think of something - has any review site tried benchmarking Trinity under Win 7 and Win 8 to see if scheduler optimisations help performance? I know the early BD tests showed moderate improvements in lightly threaded loads between Win 7 and Win 8 dev preview - with the release preview already kicking around (and therefore - presumably - better optimised) I'm surprised no-one's tested this...

Its been tested here:

http://www.planet3dnow.de/vbulletin/...#content_start

However,it is a tad pointless. Single threaded performance or performance using all four threads won't change. They needed to test situations where two to three threads are used,and the only test which appears to use two threads is Crysis. However,they use the IGP instead which means its GPU limited!!

At times it makes me think if reviewers know what they are testing for at times.

[scaryjim]

Gaming benchmarks, on a preview OS? *sigh* Do they know nothing about driver optimisation?

X264 first pass would be a more interesting one, or a "light" multi-tasking scenario. They should run two or three single-thread benchmarks simultaneously to see what happens - after all, that probably more realistically represents average usage scenarios for most users: having several single-thread operations running at once.

[CAT-THE-FIFTH]

A chap on OcUK said that one review of the FX8350 was released early and the CPU was around Core i7 2700k,but it was pulled. If this is true that is not too bad for around £160,although I suspect for lightly threaded stuff it will be closer to a Phenom II X4.

[watercooled]

Any idea where the review was? Google cache might still hold a copy if it wasn't too long ago and was up for long enough.

[CAT-THE-FIFTH]

Any idea where the review was? Google cache might still hold a copy if it wasn't too long ago and was up for long enough.

It was today but it seems that Google cache has no copy:

https://www.google.co.uk/search?hl=e....1.zc7-d5mh5PQ

[CAT-THE-FIFTH]

It looks like tommorrow has been confirmed as Vishera launch day:

http://www.xtremesystems.org/forums/...=1#post5145792

[CAT-THE-FIFTH]

FX8350 review under Linux:

http://www.phoronix.com/scan.php?pag..._vishera&num=1

"The slated retail price at launch for the FX-8350 is $195 USD. The Intel Core i5 3570K is presently retailing for around $230 and the Intel Core i7 3770K is around $330."

[CAT-THE-FIFTH]

This review is under Windows:

http://translate.google.com/translat...d-fx-8350.html

[DDY]

Any word on Opteron pricing?

[CAT-THE-FIFTH]

Any word on Opteron pricing?

Not yet. However,here is a list of the models at launch:

http://wccftech.com/amd-opteron-proc...iver-detailed/

[scaryjim]

You know, there's just something terribly intriguing about a 16-thread CPU with an 85W TDP that makes me want to run out and buy one. Utterly hopeless for modern desktop workloads, and single-thread performance will be diabolic, but it's just such a lovely concept Anyone know how many of those you can stack on a single board?

[Biscuit]

You know, there's just something terribly intriguing about a 16-thread CPU with an 85W TDP that makes me want to run out and buy one. Utterly hopeless for modern desktop workloads, and single-thread performance will be diabolic, but it's just such a lovely concept Anyone know how many of those you can stack on a single board?

still drives me a bit nuts that the modues are classed as 2 cores and not threads (as you put).

Surely it will be no different WRT stacking as BD (which was 4 IIRC)