Bookmark and Share


Gigabyte Technology, one of the leading manufacturers of mainboard in the world, has updated BIOS for its flagship AM3+ mainboard and accidentally reveled specifications of launch lineup of AMD FX-series central processing units code-named Zambezi. The family of AMD’s top-of-the-range chips will be consist of four eight-core models, one six-core chip and one quad-core product.

As noticed by VR-Zone web-site, Gigabyte’s GA-990FXA-UD7 mainboard already supports all of AMD’s FX chips due to be launched in 2011 with F4 BIOS version. In generally, Gigabyte nearly officially confirmed specifications of the chips that have been published for a number of times already, so, the BIOS update brings no surprises. Just as expected, the lineup of AMD FX-series will crowned by the FX-8150 chip, will contain four FX8100-series eight-core models, one FX-6100 six-core chip and one FX-4100 quad-core product.

Meanwhile, a number of market indicators revealed this week that the highly-anticipated microprocessors based on Bulldozer micro-architecture may become available later than expected. The mass availability of next-generation high-end desktop FX-series central processing units code-named Zambezi in September is under question mark at the moment, an industrial source told X-bit labs. Other sources indicated that the desktop chips powered by Bulldozer micro-architecture will likely be launched sometimes in the fourth quarter of 2011, more than a quarter later than the company originally expected. AMD did not comment on the information.

AMD Orochi design is the company's next-generation processor for high-end desktop (Zambezi) and server (Valencia) markets. The chip will feature up to eight processing engines, but since it is based on Bulldozer micro-architecture, those cores will be packed into four modules. Every module which will have two independent integer cores (that will share fetch, decode and L2 functionality) with dedicated schedulers, one "Flex FP" floating point unit with two 128-bit FMAC pipes with one FP scheduler. The chip will have up to 8MB L2 cache, shared 8MB L3 cache, new dual-channel DDR3 memory controller and will use HyperTransport 3.1 bus. The Zambezi chips will use new AM3+ form-factor and will require brand new platforms.

AMD did not comment on the news-story.

Tags: AMD, Gigabyte, Zambezi, Bulldozer, 32nm


Comments currently: 31
Discussion started: 09/01/11 03:56:52 PM
Latest comment: 09/06/11 12:44:50 AM
Expand all threads | Collapse all threads


Bulldozer resembles P4 design. Weak cores (low IPC), higher frequences and lot of cache (desperate way to increase performance). Who will use their FMA4 ? Waste of die space.
5 7 [Posted by: Tristan  | Date: 09/01/11 03:56:52 PM]
- collapse thread

Bulldozer resembles P4 design. Weak cores (low IPC), higher frequencies and lot of cache (desperate way to increase performance). Who will use their FMA4 ? Waste of die space.

Powerful cores(High IPC), Lower frequencies(The p4 design doesn't not resemble the Bulldozer design, nothing is double clocked)....and FMA4 will be used in Multimedia(x264, mp3, mkv, flv, you know the usual multimedia)
5 5 [Posted by: seronx  | Date: 09/01/11 04:07:03 PM]
There's not really anything we can tell about IPC based on "higher frequencies and lots of cache. Wait for real benchmarks before making such poor assumptions...

If you want to make such poor assumptions, then fine - i'll begin to make poor assumptions as well. What if BD does have about 20% lower IPC, but could clock to 6-7Ghz on air as opposed to 4.5-5Ghz for SB? Pentium 4 could have easily smoked the K8 if there were more optimizations done to the fab process and lowered power consumption at higher clocks - which, at 32nm (95W for 8 cores is nothing short of impressive! plus the other architectural power saving enhancements INCLUDING the shared architecture) is probably a lot more feasible.
2 2 [Posted by: 1771_bx  | Date: 09/01/11 08:24:26 PM]
Part of what you're saying is true, but part of it is not. The Pentium 4 suffered from lousy decoding logic, and about 50% of the time the processor was running as a scalar processor. The very long pipeline was often cited for the low IPC, while it was not entirely to blame.

Sandy Bridge is the latest version of the Pentium 4, not the Bulldozer. It's got a ton of Pentium 4 DNA in it, and this time it's being used successfully.

I agree if they had brought the Pentium 4 down to 45nm, it would have hit crazy clock speeds because of the big improvements with that process. But, with Conroe out, it's doubtful it would be enough, especially consider it would need a lot of power to generate good performance, even with the 45nm improvements Intel initiated.
4 0 [Posted by: TA152H  | Date: 09/02/11 07:47:52 AM]
Would you happen to know how long the Sandy Bridge pipeline is? I can't seem to find anything about that on the internet.
0 0 [Posted by: lol123  | Date: 09/03/11 10:45:20 AM]
the last iteration of the p4 had a 30+ stage pipeline.
bulldozer's is somewhere between 18 and 22 as far as we can make out.

the first pentium 4 had a pipeline length of 24 when previous cores had 12 stages.
Today core are around 14, so 18-22 stages isn't anywhere near as extreme as the p4's pipeline increase.

the p4 also suffered from a cache shortage, a memory bandwidth shortage and high memory latencies. none of these are a problem for the bulldozer.

all in all, bulldozer isn't at all like a p4.
0 0 [Posted by: Countess  | Date: 09/03/11 11:02:02 AM]
The first Pentium 4s (Willamette and Northwood) had a 20-stage instruction pipeline, not 24. Intel used a 31-stage instruction pipeline for the Pentium 4s based on the Prescott and Cedar Mill cores.
0 0 [Posted by: DirectXtreme  | Date: 09/05/11 03:01:32 PM]
that's discounting the decoder stages, which are included with bulldozer estimate. if you exclude those the CPU's at the time had a 10 stage pipeline.

and excluding those its probably ~18 stages for bulldozer.
0 0 [Posted by: Countess  | Date: 09/06/11 12:44:50 AM]

Bulldozer has a higher IPC then its older predecessor every 1Mhz os 30% faster then PhenomII add that the extra cores, higher clocks, improved increased cache and a full redesign on 32nm and you have a CPU that can compete with Intel at the same price range.

Thats all AMD wants and it's what AMD has and will release.

The pentium 4 is what happens when a company forgets about making CPU better and goes for making a CPU market better.. its much easier to sell a 3GHz Pentium 4 then it is a 2Ghz Athlon64 even if the 2GHz Athlon64 blew the P4 out of the water.

By the time people started to catch on to this con job intel started to bribe its partners in blocking AMD broducts and by the time people found out about that it was almost ready to release its replacement to the P4 .. the CORE CPU
5 1 [Posted by: vid_ghost  | Date: 09/01/11 06:23:19 PM]
- collapse thread

AMD says that WHOLE CPU is 35% faster than prevoius generation. So IPC per core is the same as with Phenom II, probably with few exceptions.
1 0 [Posted by: Tristan  | Date: 09/02/11 12:20:55 PM]
Maybe he was comparing 8 cores vs 8 cores... 16 vs 12 should give even more performance.
0 2 [Posted by: Filiprino  | Date: 09/02/11 05:47:16 PM]

words are useless until benches....
2 3 [Posted by: TAViX  | Date: 09/02/11 06:45:47 AM]

I want to see benchmarks ASAP!!
2 2 [Posted by: Filiprino  | Date: 09/02/11 09:55:10 AM]

Let's see: 8 cores @ 4.2 GHz = 33.6 GHz of aggregate power.

Next stop -- 5.0 GHz (cf. "unlocked" !!

Go AMD!!!

We L-U-V competition.

1 3 [Posted by: MRFS  | Date: 09/02/11 11:25:19 AM]

So the FX-8120 has a TDP of 95W or 125W?
0 0 [Posted by: eltoro200  | Date: 09/02/11 11:47:30 AM]
- collapse thread

FX-8120 will be in both modifications with different TDP.
0 0 [Posted by: Tester128  | Date: 09/02/11 02:33:58 PM]
Oh, now I see "...will contain four FX8100-series eight-core models."
FX-8100, 2 x FX-8120 and FX-8150.
0 0 [Posted by: eltoro200  | Date: 09/03/11 04:29:06 AM]

actually IPC doesn't make much difference. lets say we have 10% lower IPC but whit this compremise we could have an increase in clock of 25% than the cpu would still be faster and the cores would be smaller so more cores / mm^2 so its a double win. and if only 1 core in a module is active then you would have 2 times the FPU performance compared to a SB core.
0 1 [Posted by: massau  | Date: 09/03/11 02:18:08 PM]

I really don't see the point of buying the FX-8120 or FX-8150, when FX-8100 will have an unlocked multipiler as well.
0 0 [Posted by: SteelCity1981  | Date: 09/03/11 11:49:50 PM]
- collapse thread

but it has no garantie it will reach 3.6 or 4.2ghz. and most people do not overclock.
0 0 [Posted by: Countess  | Date: 09/04/11 09:42:07 AM]
Ok lets see that it's based on the same arch as the rest of the 8000 series but underclocked so why wouldn't it reach those speeds? It's not like the 8100 has a diff stepping then the 8150 where it couldn't reach those speeds so there for it is a gurantee. It's lower clocked because of the price sales point that's all.
0 0 [Posted by: SteelCity1981  | Date: 09/04/11 07:17:28 PM]
Chips are often tested before assigning speed bins. Another common trick is to take the highest speed bin or ULV chips from the centre of the wafer (generally higher quality) and the lower speed items from the edges. Often the lower clocked parts are identical to their higher clocked counterparts, but this is not always or necessarily the case.
0 0 [Posted by: Prosthetic_Head  | Date: 09/05/11 04:13:43 AM]
because of various techniques (real-time process adjustments) the difference in chips from the centre and chips from the edges of the wafer has pretty much disappeared.

you are correct on the binning part thou.
0 1 [Posted by: Countess  | Date: 09/05/11 05:45:10 AM]

0 0 [Posted by: tbaracu  | Date: 09/04/11 12:23:48 PM]
- collapse thread

Still not trustworthy benchmarks.
0 1 [Posted by: Pouria  | Date: 09/04/11 02:31:32 PM]
Gee and what kind of site is that again? Obv that site isn't a trusted site, because hardly anyone has heard of that site. Yeah lets wait until toms hardware does the benchmarks.
0 0 [Posted by: SteelCity1981  | Date: 09/04/11 07:20:14 PM]
The sites quotes its source as a Sisoft review that may have accidentally gotten published before NDA end. Not many saw it, but apparently Google did:

0 0 [Posted by: jihadjoe  | Date: 09/05/11 01:37:05 AM]
The figures quoted there don't seem unreasonable, but bare in mind:

The speeds quoted here (2.8GHz with turbo to 3.8GHz) doesn't match with any of the models above. It is nearest to the slowest 8xxx series model (FX 8100, 2.8-3.7GHz). The higher models should outperform these benchmark figures.

It's also worth remembering that these benchmarks, especially of pre-release chips are not a good guide to real world performance of production chips.
0 0 [Posted by: Prosthetic_Head  | Date: 09/05/11 04:36:03 AM]
those benchmarks are from a old engineering sample with pre-fetching disabled/non-functional.
0 2 [Posted by: Countess  | Date: 09/05/11 12:49:54 PM]


Add your Comment

Related news

Latest News

Wednesday, November 5, 2014

10:48 pm | LG’s Unique Ultra-Wide Curved 34” Display Finally Hits the Market. LG 34UC97 Available in the U.S. and the U.K.

Wednesday, October 8, 2014

12:52 pm | Lisa Su Appointed as New CEO of Advanced Micro Devices. Rory Read Steps Down, Lisa Su Becomes New CEO of AMD

Thursday, August 28, 2014

4:22 am | AMD Has No Plans to Reconsider Recommended Prices of Radeon R9 Graphics Cards. AMD Will Not Lower Recommended Prices of Radeon R9 Graphics Solutions

Wednesday, August 27, 2014

1:09 pm | Samsung Begins to Produce 2.13GHz 64GB DDR4 Memory Modules. Samsung Uses TSV DRAMs for 64GB DDR4 RDIMMs

Tuesday, August 26, 2014

10:41 am | AMD Quietly Reveals Third Iteration of GCN Architecture with Tonga GPU. AMD Unleashes Radeon R9 285 Graphics Cards, Tonga GPU, GCN 1.2 Architecture