AMD Zambezi news, info, fans !

[bamtan2]

unless the 6100 unlocks to be 8100, and unless 8150 is lower voltage than 8120, everybody is just going to get 8120

15% extra money is worth 30% extra cores. and higher stock speed is not worth $40 if they both overclock the exact same.

[Opteron146]

Well yes, the parts are indeed different. However, your point was it would be shipped to OEM's and retail wouldn't see any (if at all). Unless it is a OEM special, like the 960T, where AMD clarified as much, i don't have a reason to believe that the 95W chip won't be seen in retail. If you look at it, Phenom II 945 is merely locked version of 940 at lower TDP. This being FX, i don't think AMD would muck about.

Well the 945 was later in the C3 revision a plain, normal 95W part. The first 940/945 models with 125W were still C2. Nothing exciting about that, that's the normal process/stepping optimizing benefit. If you want to compare it to FX, then my point would be that the 8120 95W model will be scarce as hen's teeth in etail, because it will be an OEM only model and etail=we have to live with the 125W part or the 8100 95W model, instead. Then next year, when they'll launch 8170, there's maybe enough 95W 8120 for everybody. We'll see...

[wez]

unless the 6100 unlocks to be 8100, and unless 8150 is lower voltage than 8120, everybody is just going to get 8120

15% extra money is worth 30% extra cores. and higher stock speed is not worth $40 if they both overclock the exact same.

With the previous launches (agena/thuban) there was a notable difference in overclocking between the top and lower models. And with the rumored somewhat limited production of BD, I dont think it too far fetched to assume the top model will be the one to get for max clocks this time around as well.

And with 8 cores, the chance of getting a crap core or two is not exactly smaller than before. So you'd assume binning plays and even bigger role than before. But yea, we'll see soon enough

[Leeghoofd]

Anyone heard about Windows 8 seeming to steer the cores better than windows 7 ? Current Task scheduler seems to mess up a bit the performance of our beloved BD... Looking forward to fire it up tonite...

[dess]

Anyone heard about Windows 8 seeming to steer the cores better than windows 7 ? Current Task scheduler seems to mess up a bit the performance of our beloved BD... Looking forward to fire it up tonite...

An update to fix it is expected sometime soon.

[wez]

Anyone heard about Windows 8 seeming to steer the cores better than windows 7 ? Current Task scheduler seems to mess up a bit the performance of our beloved BD... Looking forward to fire it up tonite...

You are such a tease...

[Smartidiot89]

An update to fix it is expected sometime soon.

Really, where have you heard this? Not supposed to be "fixed" until you get Windows 8

[undone]

http://tipidpc.com/viewtopic.php?tid=172787&page=4183

[Opteron146]

Really, where have you heard this? Not supposed to be "fixed" until you get Windows 8

I assume he meant the fix will come sometime for Win7, but Win8 will have it from the start. Just guessing from the - we don't speak about - guy's statement that Win8 is logging in BD as 4core/8thread machine.

[informal]

I think we should call Zambezi 8150 an 8 threaded CPU. This way we avoid any core vs not-a-core arguments. It has 8 strong threads,according to AMD,so let's call it 8T capable CPU. Thuban is 6T chip,SB 2600k is 8T chip. Whether it's a weak or strong thread is debatable though.

On another note,we have some shops listing FX models and mobo and FX bundles. So it seems 12th is the date. Won't be long now .

[Dimitriman]

I think we should call Zambezi 8150 an 8 threaded CPU. This way we avoid any core vs not-a-core arguments. It has 8 strong threads,according to AMD,so let's call it 8T capable CPU. Thuban is 6T chip,SB 2600k is 8T chip. Whether it's a weak or strong thread is debatable though.

On another note,we have some shops listing FX models and mobo and FX bundles. So it seems 12th is the date. Won't be long now .

I just call it a 4 module. I like the term and it "implies" 8 threads (or integer units).

So 2 Cores>Module>Hyperthreading, works for me.

[FlanK3r]

oh, I like this package....

[dess]

Really, where have you heard this? Not supposed to be "fixed" until you get Windows 8

There were some talks earier that an update is needed to optimize taskscheduling for BD, and it's to come.

BTW, AFAIK there must be also an update to come to enable FMA4 and XOP under Win7 (just like AVX were enabled by SP1).

[undone]

There were some talks earier that an update is needed to optimize taskscheduling for BD, and it's to come.

BTW, AFAIK there must be also an update to come to enable FMA4 and XOP under Win7 (just like AVX were enabled by SP1).

If it's true I would assume such these patches wont be release until Zambezi shipping to the reviewers. Wait and see.....

[Particle]

Windows itself doesn't need to be aware of FMA4 and XOP for developers to use them of course. A Windows patch would only address what Windows itself uses which can increase core Windows performance.

--

Can we not get into the core vs thread thing again?

[Opteron146]

BTW, AFAIK there must be also an update to come to enable FMA4 and XOP under Win7 (just like AVX were enabled by SP1).

No, it was needed for AVX because it introduced new/wider registers. The OS has to save these between context switches now, too. Think about what would happen if the OS forgets to save your data ... *g*

XOP and FMA4 however do not introduce new registers, they use the SSE or AVX' registers, therefore no extra patch needed - as long as the AVX patch is in place everything is fine.

[undone]

No, it was needed for AVX because it introduced new/wider registers. The OS has to save these between context switches now, too. Think about what would happen if the OS forgets to save your data ... *g*

XOP and FMA4 however do not introduce new registers, they use the SSE or AVX' registers, therefore no extra patch needed - as long as the AVX patch is in place everything is fine.

How about the new register of second integer core?

[PatRaceTin]

i see bundle package in extreme news section

[tifosi]

...there's maybe enough 95W 8120 for everybody. We'll see...

Well, process is only going to improve with time. This will allow some headroom and a chip with a lower TDP is not so far fetched an idea, as you seem to propose. There were also server MC chips, some of which now are there with lower TDP.

[Opteron146]

Well, process is only going to improve with time. This will allow some headroom and a chip with a lower TDP is not so far fetched an idea, as you seem to propose. There were also server MC chips, some of which now are there with lower TDP.

Hmm exactly what I meant.
You wrote "improve with time", that's correct, hence I wrote "next year":

Then next year, when they'll launch 8170, there's maybe enough 95W 8120 for everybody. We'll see...

I don't see a problem here, only a misunderstanding ;-)

How about the new register of second integer core?

As long as these cores are only using good,old, standard-x86 registers, no problem at all ;-)

[chew*]

http://tipidpc.com/viewtopic.php?tid=172787&page=4183

Boxxes are so last year.

I want to see tray shots full of chips

[Dumo]

Boxxes are so last year.

I want to see tray shots full of chips

Does that mean we should bin retail?

[Mechanical Man]

Boxxes are so last year.

I want to see tray shots full of chips

Then show us some!

[Voodoo²]

Look at the production date of that chip. "1136" that is september right?

[dess]

No, it was needed for AVX because it introduced new/wider registers. The OS has to save these between context switches now, too. Think about what would happen if the OS forgets to save your data ... *g*

XOP and FMA4 however do not introduce new registers, they use the SSE or AVX' registers, therefore no extra patch needed - as long as the AVX patch is in place everything is fine.

I know, but it could be that perhaps some state attributes needs to be stored. Can't find where I've read about it.

Regarding task-scheduling, it's from JF-AMD:

Performance is based on:
The silicon
The microcode in the silicon
The BIOS
The compiler updates
The drivers
The OS optimizations
Performance tuning by engineers

Also, there were some slides on how Windows' scheduler needs to be changed to accomodate to BD, and IIRC it was about Win7. Can't find it now, either.