AMD 1090T BE Processor and 1055T (Thuban) tested

[cbn]

That super Pi scores seem rather off. Is that program even multi threaded??

My X2 @ 3.8 completes it in about 18 sec.

Was it because the person was using 32-bit OS ??

Super Pi is single threaded.

 

[DrMrLordX]

The problem is Core i7 920 was around since November 17, 2008 and its release price was $284 USD. So now consider that 1.5 years later, AMD is releasing a 6-core processor for $300 that still won't outperform a 4.0ghz i7 920. That's not so rozy.

The platform cost for a 920 at release was much higher than the platform cost for a 1090T today. Not that that necessarily matters, since LGA1366 has gotten a bit cheaper for board+proc, but still, thought I'd point that out.

The main thing to remember here is that Thuban is an improvement over Deneb. They've rolled out two extra cores + turbo within the same thermal envelope and at the same basic price levels. Yeah, now you can spend $300 on an AMD processor again, but you can also spend $200 on a Thuban if you don't like the 1090T's price/performance ratio.

Personally I want to see how the 1035T overclocks, but we haven't even seen how the 1090T overclocks yet, so it'll be awhile before we see numbers like that.

Thuban closes the performance gap between AMD's and Intel's offerings at price-points below $1k. This is a win for AMD, even if it is not necessarily a crushing victory. Thuban doesn't have to surpass Intel's offerings so long as it can provide AMD with a way to catch up a bit. Let's see how 4 ghz Thuban works out . . . it may surprise us all, especially if the new E0 stepping brings good NB clocks and improved RAM overclocking potential.

 

[Hyperlite]

I don't follow the logic on noncompetition @ $300 vs the i7 860. @ 3.9Ghz, it's carrying a 40% OC. Subtract, say, 30% from those benchmark scores (flawed, i know), and the 1090T is a serious competitor with much better multithreaded performance.

edit: actually i went and did the math. 1090T should only lose 2 of those benches stock vs stock. No one has any idea how the 1090T will OC, so why are we already declaring it "not so rozy?" The way i see it, this is already a success. i5 isn't even in the picture for anyone on the fence, and the i7 (at least 1156 and lower end 1366) finally has competition.

 

[aigomorla]

dude its a 300 dollar hexcore.

for 300 dollars i think its doing a dayam good job holding up.

Its not a turtle... and its not a hare... but for 300 dollars, its a good chip with a lot of expandability on a very old socket design.

While u guys are saying oh it loses to an i5 and i7, well.. socket AM2 has been around for a very long time, and its another upgrade feature to those who went AMD @ budget.

 

[VirtualLarry]

You're going to get one, and bench it for us, aren't you Aigo?

 

[richierich1212]

Don't put a lot of stock into this first review of the 1090T. The guy doesn't even have the best ram timings (which SuperPi loves), 1333 8-8-8-ish. There are still BIOS issues to be worked out and he didn't overclock it with 6 cores (or hasn't uploaded screenshots yet).

Either way, yes just as Aigo said AMD has hit the market right in the middle. Would you go with 2 cores or 6 for about the same price? And it's a drop in upgrade for many users with AM2+/AM3 mobos.

 

[DrMrLordX]

. Would you go with 2 cores or 6 for about the same price?

If those i5 Clarksdales weren't a horrible deal before, they sure are now (or will be once Thuban is officially launched).

 

[richierich1212]

updated OP with 2nd updated testing link, looks like some overclocking action now.

 

[RussianSensation]

I don't follow the logic on noncompetition @ $300 vs the i7 860. @ 3.9Ghz, it's carrying a 40% OC. Subtract, say, 30% from those benchmark scores (flawed, i know), and the 1090T is a serious competitor with much better multithreaded performance.

Good points! I was just saying the situation is not so rozy from an efficiency perspective. You are comparing a 4-core with HT virtually threaded processor to a 3.2ghz 6-core processor. I would have expected it to keep up a lot better.

Aigo, can you run these benches on a HT-disabled 980X at 3.2ghz to see what the core clock for clock advantage the i7 holds at the moment?

 

[Dark4ng3l]

http://forum.coolaler.com/showthread.php?t=235972&page=3

That thread claims the 1055t overclocks to 3.85GHz on stock air cooling? Looks like 4+ is possible with good cooling then...

 

[Hyperlite]

Good points! I was just saying the situation is not so rozy from an efficiency perspective. You are comparing a 4-core with HT virtually threaded processor to a 3.2ghz 6-core processor. I would have expected it to keep up a lot better.

Aigo, can you run these benches on a HT-disabled 980X at 3.2ghz to see what the core clock for clock advantage the i7 holds at the moment?

duuurrrr sorry for some reason i had it in my head last night when i wrote that the 860 didn't have HT. I'll edit that to say "equal or slightly better multithreaded performance."

http://forum.coolaler.com/showthread.php?t=235972&page=3

That thread claims the 1055t overclocks to 3.85GHz on stock air cooling? Looks like 4+ is possible with good cooling then...

promising! 3.85ghz @ 1.425 vcore. the SiS bench where it smashes the 980x is interesting...does that program not take advantage of hyperthreading?

edit: nvm i was just looking at the combined score...that's still impressive though, handily beats a 980X @ 3.85 ghz in SiS. that seems a bit odd to me...

 

[richierich1212]

promising! 3.85ghz @ 1.425 vcore. the SiS bench where it smashes the 980x is interesting...does that program not take advantage of hyperthreading?

edit: nvm i was just looking at the combined score...that's still impressive though, handily beats a 980X @ 3.85 ghz in SiS. that seems a bit odd to me...

Well looks like it selected base components for the 980X. 3GB, PC3-8500, HDD, 4870X2?
And the tester has a SuperTalent SSD, Nvidia 480GTX. Well considering that the 980X costs $800 more than the 1055T (hopefully), I guess you can select those components, lol.

 

[aigomorla]

You're going to get one, and bench it for us, aren't you Aigo?

i would love a MC system...

But, im done with hexcores... You guys know how long a cpu lasts in my hands..

I got word possibly beckton will hit LGA1366 on the intel side as the last upgrade into 22nm.

:hmm:

8 cores... 16t, and an unlocked multi.. :sneaky:

But that cpu will probably come out as a xeon and hold a 2k price tag ()

 

[formulav8]

AMD could care less about competing with overclocked Intel cpu's. IF the pricing is correct then they fall into the right area at stock speeds for both sides.

Obviously as a personal preference compare to a overclocked Intel cpu but hinting that AMD shouldn't be priced in a certain segment based on overclocked performance is pretty nutty. :/


Jason

 

[aigomorla]

Obviously as a personal preference compare to a overclocked Intel cpu but hinting that AMD shouldn't be priced in a certain segment based on overclocked performance is pretty nutty. :/

2 flips of a coin here pal..

If they didnt want to compare oc potential... they wouldnt of made it a BE.. :\

I dont know about you, but when i see the words "Black Edition"... it means Overclocking to the MAX!

 

[solofly]

http://vr-zone.com/articles/amd-phenom-ii-x6-1055t-hits-retail-in-china-pictured/8804.html

Notice, the 2 kilo heatsink isn't bolted down...
http://translate.google.com/transla...ew.com/10081.html&sl=zh-CN&tl=en&hl=&ie=UTF-8

 

[Hyperlite]

stock cooler looks nice

 

[richierich1212]

http://forum.coolaler.com/showthread.php?t=236011&page=6

1055T @ 4.2GHz / NB @ 2.7GHz hits 7.38 in Cinebench 11.5. Nice.

 

[thilanliyan]

^That's a lot of voltage though...if CPUz is reading correctly.

 

[richierich1212]

^That's a lot of voltage though...if CPUz is reading correctly.

He said the voltages are reading incorrectly.

 

[thilanliyan]

He said the voltages are reading incorrectly.

Ah okay.

 

[Dark4ng3l]

Ah okay.

His other monitoring program shows 1.475v instead of the 1.568 reported by CPU-Z.

 

[thilanliyan]

His other monitoring program shows 1.475v instead of the 1.568 reported by CPU-Z.

That's a lot better than I expected actually at 4.2GHz.

 

[Accord99]

CoreTemp generally doesn't read the actual voltage, only the default voltage specified for that particular voltage (probably in this case the turbo voltage). I'd expect CPU-Z to be more accurate.

 

[Hyperlite]

CoreTemp generally doesn't read the actual voltage, only the default voltage specified for that particular voltage (probably in this case the turbo voltage). I'd expect CPU-Z to be more accurate.

I agree. CoreTemp has never shown my voltage correctly, and CPU-z has always been accurate. I'd be more inclined to think he really is putting 1.55v into it.