Thuban & Zosma info/code name from anandtech "Juicy"

[[XC] gomeler]

Thuban is too late... unless given one I don't see myself testing/benching it. If Thuban had been available say 4-5 months ago it would have turned some heads. But by May-June we'll all be in Intel lala land with Westmere. We needed the dozer yesterday. Even if they manage to release a 3GHz Thuban in the 125w power envelope the Core i7 920 and Core i7 860 will eat it alive in performance and quite possibly power consumption. There is no way they can let a 6 core die go for less than $200 so they can't compete on price and I doubt they can get clockspeeds up to 3.4GHz to compete on performance. It is bleak till 2011.. Diablo 3 will be out before Bulldozer

edit: I don't suppose there is a chance at a 12 threaded Thuban MCM at 3 jiggahurts at 140w? Using low leakage binned Thubans? I guess they'd have to beat on that 45nm SOI like it owes them money in order to keep the TDP in an acceptable realm.

[Florinmocanu]

Thuban is too late... unless given one I don't see myself testing/benching it. If Thuban had been available say 4-5 months ago it would have turned some heads. But by May-June we'll all be in Intel lala land with Westmere. We needed the dozer yesterday. Even if they manage to release a 3GHz Thuban in the 125w power envelope the Core i7 920 and Core i7 860 will eat it alive in performance and quite possibly power consumption. There is no way they can let a 6 core die go for less than $200 so they can't compete on price and I doubt they can get clockspeeds up to 3.4GHz to compete on performance. It is bleak till 2011.. Diablo 3 will be out before Bulldozer

edit: I don't suppose there is a chance at a 12 threaded Thuban MCM at 3 jiggahurts at 140w? Using low leakage binned Thubans? I guess they'd have to beat on that 45nm SOI like it owes them money in order to keep the TDP in an acceptable realm.

Dude, HT on 920/860 is around 15-20% extra performance in multithreaded apps.

a 6 core Thuban at 3ghz will beat the 920/860 (they revolve around 16-17k in cinebench 10)

I suspect around 20k in Cinebench 10, which is around what a I7 975 gets out.

a 6 core thuban at 3ghz will be a serious competitor in multithreaded apps.

Only the westmere cpus will be able to beat it, but at a much higher price tag.

[accord99]

Dude, HT on 920/860 is around 15-20% extra performance in multithreaded apps.

But then each core on the i7 is also more powerful.

a 6 core Thuban at 3ghz will beat the 920/860 (they revolve around 16-17k in cinebench 10)

I suspect around 20k in Cinebench 10, which is around what a I7 975 gets out.

It'll be more like 17K.

[Florinmocanu]

But then each core on the i7 is also more powerful.


It'll be more like 17K.

Nope, your wrong.

a Phenom II 940 scores around 12.5k in R10 Cinebench.

so that means an average of around 18.2k for a 6 core at 3ghz, but you also have higher HT/NB speeds because 965 and 955 brought increases in that area plus higher DDR3 speeds, which also help in 3d rendering.

Overall, a 6 core 3ghz CPU should score close to 19k, which is around a I7 965 XE. If we OC it to 4 ghz a 25k result should be quite easy to get out, higher than a 4ghz I7 which scores around 22.5k (mine scores around that figure).

a Thuban will be a powerfull solution for users who needs lots of rendering power, folding power etc.

[[XC] gomeler]

Dude, HT on 920/860 is around 15-20% extra performance in multithreaded apps.

a 6 core Thuban at 3ghz will beat the 920/860 (they revolve around 16-17k in cinebench 10)

I suspect around 20k in Cinebench 10, which is around what a I7 975 gets out.

a 6 core thuban at 3ghz will be a serious competitor in multithreaded apps.

Only the westmere cpus will be able to beat it, but at a much higher price tag.

Are you forgetting the fact that the 920/860 will do 4GHz easy? C3 965s aren't as hot as C2s but they are still hot enough that adding 2 more cores makes me wonder if you'll be able to not be thermally limited in overclocking. I guess we'll see what happens when the chips become available

[Lokinhow]

Are you forgetting the fact that the 920/860 will do 4GHz easy? C3 965s aren't as hot as C2s but they are still hot enough that adding 2 more cores makes me wonder if you'll be able to not be thermally limited in overclocking. I guess we'll see what happens when the chips become available

but thuban isn't C3...

[initialised]

Are you forgetting the fact that the 920/860 will do 4GHz easy? C3 965s aren't as hot as C2s but they are still hot enough that adding 2 more cores makes me wonder if you'll be able to not be thermally limited in overclocking. I guess we'll see what happens when the chips become available

I did a C3 4GHz on water under 60C a while back, with that kind of thermal performance 4GHz x6 should be possible.

If we were to compare the R10 results for AMD 965 and Intels i5 750 one will notice that the scaling is better in 965 than in i5 750.

PhII 965 does 3941 for single and 14012 for multi
i5 750 does 4238 for single and 14142 for multi
i7 920 does 3846 for single and 16211 for multi

In i7's case HT helps out a lot..

4GHz C3 965 18k x3.6
4GHz i7 23k x4
Hexacore:
Stock 24.7k
3.6GHz 28k
4GHz 31.9k
Scaling factor around x5.7 regardless of clock speed so HT doesn't help as much as on the quads which typically get ~4x expect PhII x6 to be closer to 5x.

I'll have to wait and see how the Thuban x6 goes when the sample comes in but it does have a tough act to follow given how easy 4GHz has been on i7 compared to PhII and the Clock-Clock performance discrepancy.

[Florinmocanu]

I did a C3 4GHz on water under 60C a while back, with that kind of thermal performance 4GHz x6 should be possible.



4GHz C3 965 18k x3.6
4GHz i7 23k x4
Hexacore:
Stock 24.7k
3.6GHz 28k
4GHz 31.9k
Scaling factor around x5.7 regardless of clock speed so HT doesn't help as much as on the quads which typically get ~4x expect PhII x6 to be closer to 5x.

I'll have to wait and see how the Thuban x6 goes when the sample comes in but it does have a tough act to follow given how easy 4GHz has been on i7 compared to PhII and the Clock-Clock performance discrepancy.

Your wrong, the good scaling in I7 CPU's is due to HT, without HT we would see lower scaling than on a PhII, like in a I5 750.

And also, you calculate scaling based on single thread performance, not on the ratio between increase in frequency over increase in performance.

[initialised]

Your wrong, the good scaling in I7 CPU's is due to HT, without HT we would see lower scaling than on a PhII, like in a I5 750.

And also, you calculate scaling based on single thread performance, not on the ratio between increase in frequency over increase in performance.

That's data I took on a Gulftown, not an assumption.



What I was pointing out was that the linear 4x scaling on i7 quads does not extend to hexas, it appears to drop as core count goes up.

[[XC] gomeler]

That's data I took on a Gulftown, not an assumption.

What I was pointing out was that the linear 4x scaling on i7 quads does not extend to hexas, it appears to drop as core count goes up.

Depends completely on the application at hand.

AMD should have focused on getting 32nm out of the door to allow high clock speeds and less TDP, because a CPU at 140W just isn't cutting it. I love AMD and I'm curious about these processors, but I don't like seeing a product using so much power.

AMD is a lot larger than 2 dudes in a garage banging out processors and working on fabrication. They have multiple CPU design teams, some working on Bulldozer while others working on K10.5 derived parts like Istanbul/Thuban. Global Foundaries is the company working on 32nm fabrication, AMD can't do anything until their 32nm node is functional.

[FlanK3r]

Thuban is too late... unless given one I don't see myself testing/benching it. If Thuban had been available say 4-5 months ago it would have turned some heads. But by May-June we'll all be in Intel lala land with Westmere. We needed the dozer yesterday. Even if they manage to release a 3GHz Thuban in the 125w power envelope the Core i7 920 and Core i7 860 will eat it alive in performance and quite possibly power consumption. There is no way they can let a 6 core die go for less than $200 so they can't compete on price and I doubt they can get clockspeeds up to 3.4GHz to compete on performance. It is bleak till 2011.. Diablo 3 will be out before Bulldozer

edit: I don't suppose there is a chance at a 12 threaded Thuban MCM at 3 jiggahurts at 140w? Using low leakage binned Thubans? I guess they'd have to beat on that 45nm SOI like it owes them money in order to keep the TDP in an acceptable realm.

WTF? Not, i dont mean, Westmere is something special. Performance as Athlon X3/X4, only good TDP and OC potencial (for extreme is the same as Phenoms II: 6200-6900 MHz most). And price = ? Intel will launch only Gulftown +- at the same time with Thuban form AMD. Nothing quad Westmere at 1H 2010. Launch Thubans will April/May date. TDP is lower, because CPUis at new silicon die. And i said, first Sandy Bridge will whorse performance than Gulftown. Dont fear for AMD