-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD K8L: Quad Core Desktops And So Much More
- Thread starter Azza
- Start date
yeah, read this on anandtech. looks like the memory controller will be flexible if it can handle DDR2-3 and FB
quad core with diff voltages for each one would be great too. especially if right now one core can clock higher than the other but needs more voltage
quad core with diff voltages for each one would be great too. especially if right now one core can clock higher than the other but needs more voltage
Soldato
- Joined
- 1 Sep 2005
- Posts
- 10,001
- Location
- Scottish Highlands
Is there any confirmation of what Socket the K8L will be on?
Associate
- Joined
- 29 Dec 2005
- Posts
- 1,195
does that mean that we will have to upgrade the boards again or will the boards support DDR3
Except the link above suggests that K8L will be first used in the Quad Core "Greyhound" , A) HT will be upgraded from HT2, to HT3, which has a max frequency of 2.6Ghz instead of 1.4. B) It may also have additional HT channels to allow for external Co'Processors, and C) it may also support DDR3.
All of those possible changes could mean a new socket will be required.
Perhaps AMD will shoehorn a dualcore K8L into AM2, but its all just speculation at this point.
The K8L is still rumored to be a server chip, so probably branded Opteron, so what plans they may have for K8L desktop chips is unknown.
All of those possible changes could mean a new socket will be required.
Perhaps AMD will shoehorn a dualcore K8L into AM2, but its all just speculation at this point.
The K8L is still rumored to be a server chip, so probably branded Opteron, so what plans they may have for K8L desktop chips is unknown.
Last edited:
Associate
- Joined
- 15 May 2004
- Posts
- 492
I've heard some other stuff from what I think are more reliable sources.
Firstly we're looking at a 1H07 release, properly available by the summer. It's supposed to be for all AMD chips, including replacement of the K8 on the desktop.
DDR2 and DDR3 will not be supported together. DDR3 will come later. However, DDR3 and FB-DIMMS will be supported by the same processors.
1GB page size. Yes, one gigabyte. It'll also allow far smaller page sizes, but 1GB will help ensure the K8L ends up in quite a few super computers.
Expandable L3 cache. Possibly expandable through Z-RAM.
Some new instructions, perhaps SSE4.
Two 128 bit SSE instructions per clock.
All in, on paper it sounds like being far stronger in FP than the Core2 and equivalent in most other things. This will be a more than worthy competitor to Conroe... and of course is quad-core.
Some of the stuff I mentioned has been confirmed, some strongly hinted at and some is supposition. Bottom line is that AMD didn't sit about doing nothing for two and a half years like people seemed to think they were. They were not caught off guard by Conroe. AM2 is mostly just a hack for cheaper RAM and some slight improvements. It'll ramp up with the process shrink and be competitive enough for AMD chips to still be bought. And then K8L will take back the performance crown.
It's like ATI and nVidia, except instead of switching positions every few months it's every six months to a year.
All I know is that Rosetta@Home really should fly on a K8L
Firstly we're looking at a 1H07 release, properly available by the summer. It's supposed to be for all AMD chips, including replacement of the K8 on the desktop.
DDR2 and DDR3 will not be supported together. DDR3 will come later. However, DDR3 and FB-DIMMS will be supported by the same processors.
1GB page size. Yes, one gigabyte. It'll also allow far smaller page sizes, but 1GB will help ensure the K8L ends up in quite a few super computers.
Expandable L3 cache. Possibly expandable through Z-RAM.
Some new instructions, perhaps SSE4.
Two 128 bit SSE instructions per clock.
All in, on paper it sounds like being far stronger in FP than the Core2 and equivalent in most other things. This will be a more than worthy competitor to Conroe... and of course is quad-core.
Some of the stuff I mentioned has been confirmed, some strongly hinted at and some is supposition. Bottom line is that AMD didn't sit about doing nothing for two and a half years like people seemed to think they were. They were not caught off guard by Conroe. AM2 is mostly just a hack for cheaper RAM and some slight improvements. It'll ramp up with the process shrink and be competitive enough for AMD chips to still be bought. And then K8L will take back the performance crown.
It's like ATI and nVidia, except instead of switching positions every few months it's every six months to a year.
All I know is that Rosetta@Home really should fly on a K8L
yeah some good points there, everybody is talking about conroe dominating the current incarnation of the K8 processor, AM2 is coming soon, and yes i know it looks bad for now, but im sure it'll improve when they switch to 65nm later this year i think and DDR2 timings get tighter, i think the 20% lead conroe has will shrink rather than increase when its released. look what they did with the newer venice and san diego variants of the K8, added new instruction sets, improved the silicon, improved there thermal output, i don't doubt this will happen with AM2 as well and who knows what the story will be when conroe eventually hits the shelves in the summer, it might only have 5 - 10% performance lead over AM2. if thats the case AMD won't loose too much this round and can concentrate on keeping AM2 competative and getting K8L on the streets ahead of schedule while keeping AM2 upto date and revised every now and again, intel haven't bought me over just yet, waiting to see what the boys at AMD can come up with before jumping on the conroe/merom boat like most others already have
Associate
- Joined
- 15 May 2004
- Posts
- 492
AM2 will probably not bring anything significant to the table. It'll be similar performance wise to the current chips. The first process shrink will be the same core on a smaller process, in the way that the Winchester was merely the Newcastle core shrunken. Venice and San Diego were natively designed for 90 nm and brought a little bit in terms of performance. But the main thing was they go faster. The fastest Opteron is 3 GHz, and we've seen how many Venices and San Diegos will go to speeds pushing that number. That is one of the main reasons for a process shrink.
Having had a discussion about it on Ars Technica, the consensus was that 3 GHz would be breached and would probably be ramped to nearly 4 GHz. That would mean that even with Conroe having a 20% clock for clock performance advantage, that AMD would still at least be competitive.
It's like the last days of the Athlon XP. The Northwood Cs, the 800 MHz FSB versions with Hyper-Threading, were faster than the Athlon XPs. But AMD kept the gap narrow enough that they were still competitive, even if they weren't the fastest. Then A64 arrived and CPUs changed for good. All AMD need to do is remain competitive throughout Conroe and hope the K8L is as awesome as it sounds on paper.
I'm not a fanboy. I don't care who is the fastest, I won't rule anything out. I'm just glad that there is being some advancement in terms of speed. After all, it's been 3 and a half years since the first 3 GHz P4 with HT. We've not really seen much movement from Intel since then, a grand total of 740 MHz, albeit with more cache and a faster FSB.
Having had a discussion about it on Ars Technica, the consensus was that 3 GHz would be breached and would probably be ramped to nearly 4 GHz. That would mean that even with Conroe having a 20% clock for clock performance advantage, that AMD would still at least be competitive.
It's like the last days of the Athlon XP. The Northwood Cs, the 800 MHz FSB versions with Hyper-Threading, were faster than the Athlon XPs. But AMD kept the gap narrow enough that they were still competitive, even if they weren't the fastest. Then A64 arrived and CPUs changed for good. All AMD need to do is remain competitive throughout Conroe and hope the K8L is as awesome as it sounds on paper.
I'm not a fanboy. I don't care who is the fastest, I won't rule anything out. I'm just glad that there is being some advancement in terms of speed. After all, it's been 3 and a half years since the first 3 GHz P4 with HT. We've not really seen much movement from Intel since then, a grand total of 740 MHz, albeit with more cache and a faster FSB.