AMD Phenom II X6 1055T & 1090T Reviews

[drmrlordx]

Are the 960T and 940T 6 cores with 2 cores disabled, and if so, will or is AMD making any native 4 core dies, with these new cores?

The 960T and 940T will be Zosma chips, which are nothing more than Thubans with two failed (locked/disabled) cores. Word is that Zosma is unlockable in the same manner that some X2 and X3 chips were unlockable (hell even the Sempron 140 is unlockable). YMMV of course.

"Native" E0-stepping quads with turbo and other Thuban-like enhancements (such as they are) haven't yet been mentioned by anyone of which I am aware.

[Ghostbuster]

right.....

Cherry pickin' on one bench? Try the rest..







Those are pretty much fully multithreaded benches.

[wuttz]

Cherry pickin' on one bench? Try the rest..
Those are pretty much fully multithreaded benches.

cherry pick one, two, three, ten benches-
point is facts arent anywhere near xbit's rant.

[informal]

@ Ghostbuster
Your charts show 1090T in very favorable light... It loses mostly to 2 highest clocked QC and six core i7s...

[BoredByLife]

@ Ghostbuster
Your charts show 1090T in very favorable light... It loses mostly to 2 highest clocked QC and six core i7s...

Because those apps are pure mulithreaded and make use of all cores. Not many apps can do that.

[kl0012]

1090T is 3.2Ghz stock and the Intel 3.2ghz stock i7 is well more than triple the price.

This argument is pretty useles in overclocking community.

Besides most tests in the list are synthetic the only two tests that do matter are winrar and x264. As we already know winrar does not take much advantage of the added cores and as for the x264 the difference is very little.

Yeh, the diff is very little in a such multithreaded test with almost perfect multi-core scaling. Unfortunelly there are much more real-world apps which has no such perfect multicore scaling.

BTW The games are on Medium Quality..

If A game is GPU limited, then wont be much diff, but... Is there any tests with 480GTX out there?

right...

Something is definetely wrong with their test. core i7-980X is slower then i5-750 and even slower then athlon x4 630.
Here is another one:
http://www.computerbase.de/artikel/p...pt_h264avc_pro

[Ghostbuster]

cherry pick one, two, three, ten benches-
point is facts arent anywhere near xbit's rant.

Rant?

@ Ghostbuster
Your charts show 1090T in very favorable light... It loses mostly to 2 highest clocked QC and six core i7s...

Noted. Unless AMD improves the IPC per core performance, then it will continue to linger behind having to use more cores to match Intel's offerings. This is part of AMD's 'outcore' the competition, though it can only get them this far only. Just look at how far the AMD's competitor six core is at..

[superrugal]

Overclocking-wise, pretty well sumarized :

1090T NB-Frequency is default?

[Ghostbuster]

If A game is GPU limited, then wont be much diff, but... Is there any tests with 480GTX out there?

Found one here at THG -> http://www.tomshardware.com/reviews/...fx,2613-9.html

Here is another one:
http://www.computerbase.de/artikel/p...pt_h264avc_pro

Here also http://www.bit-tech.net/hardware/cpu...lack-edition/5

[informal]

Noted. Unless AMD improves the IPC per core performance, then it will continue to linger behind having to use more cores to match Intel's offerings. This is part of AMD's 'outcore' the competition, though it can only get them this far only. Just look at how far the AMD's competitor six core is at..

There is no free lunch.While intel invested in R&D and upped the transistor count per core(SMT and other improvements) ,AMD used the tweaked 10h core and upped the core count.2 different strategies,each has pros and cons. Either you invest in core logic and have larger and faster (IPC) cores,or you invest in smaller cores but cram more of them inside a chip. AMD will do similar thing intel did,next year with BD modules,upping the IPC and core count at the same time though.

[bloodbanger]

Bit dissapointed with this .Planning on buying new rig soon,had planned for an I7 920 with a Gigabyte X58 UD5 board...The first previews about the 1090 made me rethink the planning,so wanted to buy a 1090 with a Gigabyte 890FX UD 7 board.Prices are 482 euro's for the AMD way,489 for the Intel way...Easy choice i think...

[Ghostbuster]

There is no free lunch.While intel invested in R&D and upped the transistor count per core(SMT and other improvements) ,AMD used the tweaked 10h core and upped the core count.2 different strategies,each has pros and cons. Either you invest in core logic and have larger and faster (IPC) cores,or you invest in smaller cores but cram more of them inside a chip. AMD will do similar thing intel did,next year with BD modules,upping the IPC and core count at the same time though.

Upping the IPC and reintroducing SMT have greater effect on desktop computing as can be seen from Intel's progress, while having more cores would be more suited to "many-cores" and server computing. X-bit and Bit-tech not long ago compared the 6-core Istanbul to Core i7 and Nehalem Xeons, already very much predicted the outcome. Turbo core was the only way to match some of the deficiencies of the original Istanbul, though often looking at some of the benchmark data Thuban seems to perform more like a 3+ core rather than a quad one. As BD concerned, JF-AMD's comments and some pointers here (from ex-engineers) does not sound like much IPC improvement, in my view anyway. Have to wait and see.

[Florinmocanu]

The 3dsmax bench in the anandtech review,




Is not 3d rendering, it's viewport performance. Those guys at anandtech don't even know what they are testing. It's really annoying for me as a guy who does this for a living to see such irelevant benchmarks included by really highly regarded websites.

Cinebench 11.5 is the most relevant at the moment for 3d rendering, along with pov-ray. Look at those 2 benchmarks, if you want to know if that is a good rendering CPU or not.

PS: i'm a 3d professional, if anyone wants to know why i said this.

[superrugal]

The 3dsmax bench in the anandtech review,




Is not 3d rendering, it's viewport performance. Those guys at anandtech don't even know what they are testing. It's really annoying for me as a guy who does this for a living to see such irelevant benchmarks included by really highly regarded websites.

Cinebench 11.5 is the most relevant at the moment for 3d rendering, along with pov-ray. Look at those 2 benchmarks, if you want to know if that is a good rendering CPU or not.

PS: i'm a 3d professional, if anyone wants to know why i said this.

Agree. Cinebench is the most accurate benchmark I've ever seen. I have concerned about the image rendering ability of CPU since a long time ago, and found that cinebench is the best benchmarking software I can refer to.

[flippin_waffles]

Are you saying that we should all look only at benchmarks that use the extra two cores ? Why ?

I say we should look at all real world benchmarks, not only the ones that favor Phenom II X6. And as you noticed, i7 performs better in many real world applications such as winrar or photoshop and gaming. Should we ignore these real world app just because it not very well optimized for six cores ?

And yet, ironically, you want everyone to focus on the reviews that slant in intels favor. For every review that shows one thing, there is another that doesn't. I say f##k the bs reviews and their agenda, I've seen more than most of them combined with the handful of users that actually bought the chips with their own money. I trust them more than reviews from websites anymore. LostCircuits being one of a couple exceptions.

Anyway, yeah it looks like the price tag is certainly justified.

[Gilgamesh]

The 3dsmax bench in the anandtech review,

Is not 3d rendering, it's viewport performance. Those guys at anandtech don't even know what they are testing. It's really annoying for me as a guy who does this for a living to see such irelevant benchmarks included by really highly regarded websites.

Cinebench 11.5 is the most relevant at the moment for 3d rendering, along with pov-ray. Look at those 2 benchmarks, if you want to know if that is a good rendering CPU or not.

PS: i'm a 3d professional, if anyone wants to know why i said this.

Bench 3dmax10 from Pcworld.fr and Hardware.fr:

http://www.pcworld.fr/article/phenom...-excel/487121/



http://www.hardware.fr/articles/789-...90t-1055t.html



1090T on pair with i7 975 not i5 750!

[Ghostbuster]

And yet, ironically, you want everyone to focus on the reviews that slant in intels favor. For every review that shows one thing, there is another that doesn't. I say f##k the bs reviews and their agenda, I've seen more than most of them combined with the handful of users that actually bought the chips with their own money. I trust them more than reviews from websites anymore. LostCircuits being one of a couple exceptions.

And what happens if LostCircuits did not publish a favorable review? Another 'f##k the bs reviews and their agenda' rant?

[Ghostbuster]

Bench 3dmax10 from Pcworld.fr and Hardware.fr:

http://www.pcworld.fr/article/phenom...-excel/487121/



http://www.hardware.fr/articles/789-...90t-1055t.html



1090T on pair with i7 975 not i5 750!

I think that was the 1055T, not the faster 1090T..

[ajaidev]

1090T got best value and i7 980x got best performance nice and true....

[Olivon]

And what happens if LostCircuits did not publish a favorable review? Another 'f##k the bs reviews and their agenda' rant?

Those hexacores are really cool ! Always good to see some competition

[Gilgamesh]

I think that was the 1055T, not the faster 1090T..

On Anandtech's review, 1090T just better i5 750 with 3dmax (13.7 vs 13.4)

I'm reported benchs where 1090T = i7 975X or better

[Ghostbuster]

Those hexacores are really cool ! Always good to see some competition

Yups, love to see those prices fall. Affordable high performance on the desktop. Though any reduction in pricing from Intel side will not be good for AMD as the price squeeze comes down harder resulting in much lower ASPs. Just look at where the Phenom II X4s are..

[AAbenson]

hope to see Movieman commenting on this new AMD hexcore...
also some crunch results would be great

[ajaidev]

Two beauties "Bloomfield is too old everyone has seen its die"

Gulftown and Thuban



Both are so pretty

Edit: Forgot to include Thuban is 106mm2 larger than the Gulftown because its 45nm and Gulftown has 266M more transistor's than Thuban.

[OhNoes!]

emmmm, I was confused about these strange result. Very strange.

Nothing strange amigo, that is a summary of how the processors fared in the suite of tests.

Overclocking-wise, pretty well sumarized :





1.5 years later old, good i7-920 (which also can do 4.0 GHz) is still an upper-mainstram king.

Thanks for posting this, this vindicates my choice of the i7 930 (which I bought at the Microcenter for $200 only ). Don't get me wrong, AMD has to compete with what it's got, but a six-legged pig is still a pig. Even though they have made some remarkable strides with the power consumption, and introduced turbo, the deneb cores just don't cut it. Ironically, folks were arguing about quad-core deneb vs bloomfield last year, now with the intro of hexa-core deneb, we can actually see the process/ipc advantage Intel has over AMD. For my wallet's sake, I hope bulldozer brings some more ooomph. For now, the i7 920/930 is king in price/performance if one takes into account all the real world stuff; encoding, compression, rendering, photoshop, excel, gaming, etc.