i wish intel would make a tri-core
of course they can't, given their MCM quad-core design.
i wish intel would make a tri-core
of course they can't, given their MCM quad-core design.
What makes you think they can't? One chip dual core, another chip single core....Originally Posted by hollo
i wish intel would make a tri-core
of course they can't, given their MCM quad-core design.
They won't do this though, they will simply drop quad prices down to the Tri-core level... or at least that is what I would suspect.
My E6300 @ 3ghz in the CINEBENCH R10 did 5979
mm cheaper quads....but i dont think they'll need to as the performance advantage is still with the intel quad....They won't do this though, they will simply drop quad prices down to the Tri-core level... or at least that is what I would suspect.
but as intel seem to have every other niche filled....how about cheaper 9300/9450's
Last edited by adamsleath; 11-23-2007 at 08:06 PM.
i7 3610QM 1.2-3.2GHz
that's called marketing.
I am sure you do not have the fastest cpu and best mobo with graphics you're self, so what do you have to say then you bought something bad?
it's all about price/performance/features if there was no chooise like in the past we would still pay more then 2x the price.
The use and budget of you're system is most important. most don't care about max performance and best oc because they don't have the knowledge to maximize it, just bought for a friend a 5000 black with 790x and 3850, now that is value for money for a medium gamer and for this comby is easy to oc for him, he just needed a different multi to get it to 3.2 with ease, no troubles with possible htt/fsb change or memory change, just default cas4 ddr2 800.
I use 100% vmware (job) so i am happy that my system is an amd, because on medium-havy load my system performs 15-25% faster then an equal clocked c2d and global response is much better due to memory, now that is a fact and phenom/barcelona will only increase the difference.
Last edited by duploxxx; 11-23-2007 at 10:36 PM.
But it's not. In VMMark, 2 socket Clovertowns match the performance of 4 socket Opterons and Harpertowns are faster still. And when price/performance (to say nothing about reduced footprint and lower power consumption) is factored, it becomes a massacre for Intel.
And AMD still lacks the confidence to publish any meaningful Barcelona scores.
Last edited by accord99; 11-23-2007 at 10:49 PM.
The reason I am thinking this way is because ultimately Intel has a more cost effective model, their quad cores will costs less to manufacture than AMD's tri-cores.mm cheaper quads....but i dont think they'll need to as the performance advantage is still with the intel quad....
but as intel seem to have every other niche filled....how about cheaper 9300/9450's
If Intel is really wanting to put the hurt on AMD, they will drop quad core prices down to Tri-core price points, balanced with performance that is...
The jury is still out on whether the tri-core idea is good or bad, it may end up backfiring on AMD... much depending on Intel's response.
You cannot disable a core while running windows with the Phenom, at least not that I know of, but what you can do is lowering the multis on each core individually, so you can run one core at full speed and the other three at a very low one, saving energy. At least in theory. AMD Overdrive didn't really do what it was supposed to do when I tested it.
オタク
"Perfection is a state you should always try to attain, yet one you can never reach." - me =)
Phenom X3 will be competing with Wolfdale. In performance and in price. I think that has to be obvious.
I am not so sure, we don't know what Intel will do in the face of the X3 .... X3 will compete in price/performance against anyting within that price/performance range.... be it a dual core or a quad core.
In single threaded code, it will lose to wolfdale, in multithreaded code it probably would beat wolfdale (dual core). I am not so certain Intel would simply not populate the price points of the X3 with Quads that perform equal or slightly better in single threaded and whollop it in multithreaded.
Again, at a die size of 214 mm^2 compared to 283 mm^2, Intel has much more pricing power than AMD does.
You can specify the number of usable cores in the boot.ini file.
Here is the procedure, not in my words, just googled.
Of course this is at the OS level, not sure how programs/OS communicate to schedule threads on cores if the OS is only 'aware' of one, two, or three cores.You need to right click on "My Computer" -> Properties -> Advanced -> Startup & Recovery;
There, you will find a button named "edit" - by pressing it you can edit the Windows Boot.ini file manually.
at the end of the line add the following code:
/numproc=1
save the Boot.ini file and reboot
Your Windows XP Install will now only use one Processor Core.
The games which I had problems with before are now fully functional.
This site also did a X2, X3, and X4: http://www.hardware.fr/articles/694-...enom-9600.html
However the french tanslation of how they did it was not clear:
In addition to the Phenom 9600 in its original configuration, we also tested by activating only 2 then 3 core. In Configuration 2 core, this allows us to better understand the performance enhancements related to architectural changes
Last edited by JumpingJack; 11-24-2007 at 01:37 AM.
it is only simulated test, one core is off ...?
I don't know... I have never tried it (using Boot.ini) ... I will, to see what it does... not tonight, it is getting late.
What you can do is set the processor affinity in the task manager as well, this will schedule only threads on that process that is running.
Try this...
1. Run Cinbench R10, the multiCPU test as normal.
2. Open task manager. Go to the processes tab, and find the executable for Cinbench R10, right click and select Set Affinity.
3. For dual core you can select 1 or 2 processors (checked), in this case check just one, say CPU 0.
4. Run the multiCPU bench again.... it will score and thread the same as if you ran the single CPU bench.
I have done this on two dual core systems for both Cinebench and SuperPI 1M, interstingly the scores are always lower for CPU 0 than for CPU 1, I wrote that off as a scheduler issue with Windows, where some background processes only use the home node.
Jack
And how do you think 107mm^2 would fare, then. Two 3.16 GHz cores vs three 2.3 GHz ones. 6.32 Intel GHz vs 6.9 AMD ones (I know it's not such an easy calculation, but it's not that far off, either).
Just think how much it would hurt AMD's pride.
And just don't get started on overclocking.
Yes, title is:
I don't speak german, but it's enough clear without translation.Da sich beim AMD Phenom einzelne Kerne der CPU deaktivieren lassen, haben wir sowohl einen Dual-Core-Phenom als auch einen Triple-Core-Phenom simuliert und mit einem Quad-Core-Phenom und einem Athlon 64 X2 4400+ verglichen.
Well, when I stated the die size I was effectively taking 107 mm^2 x 2 which is the MCM total aggregate die size of Yorkfield (Intel upcoming 45 nm quad) AMD has already released the 283 mm^2 for their quad, and a tir core will be the same die size (one core simply disabled).
It would be inappropriate to simply additively add up clock speed to make a relative compare, because in doing so you are assuming perfect scaling for a singular app broken into multiple threads -- this would violate Amdahl's Law.
So I suspect, for most multithreaded apps a Tri-Core AMD 2.3 Ghz K10 core would beat a 3.0 Ghz C2D Wolfdale dual core... this is just a guess.
The success of tri-core against Intel dual core will depend on how the effective consumer desktop software landscape is at the time. With enough apps taking good advantage of tri-core, it could prove to be lucrative.
My concern about this approach (and honestly, I do not see any other way for AMD since they insisted on a monolithic design) is that it throws a kink into the price/performance curve, and puts out another price war so to speak.
In that, Intel could lower the price of their quads such that they win on the multi threaded performance front for the price against Tri-core -- what this does is effectively wipe out AMD's quad core offering right off the bat. AMD would need to lower quad prices, and consquently tri-core prices and it is a death spiral.
I know what you are trying to say, but according to these results (courtesy of legitreviews.com) Intel doesn't need a cheap Quad Core to compete with AMD 3Cores. At 3.33GHz, Wolfdale is much closer to full blown Quad Core Phenom, than to its 'crippled' 3 core brother.
A Wolfdale at 3.33ghz will be closer to full blow a Intel quad core at 2300mhz also, so your point is?
Well, you bring in a good point.... but a 3.33 GHz wolfdale is likely a top (or near top bin) dual core. I just don't have enough data nor a good enough sense of scaling to figure where the tri-core will fit within the performance curve to make a good estimate.
If I follow, you are arguing that Intel will have enough power/pricing in their dual core line up to compete with the tri-cores, you might be right I cannot be certain.
Jack
So ?
To be honest nobody can predict on this thread how a tricore will suceed $$ wise as there is not enough expertise here. I don't think AMD or Intel know either, so we lot have no chance.
It does give AMD more flexibility, that is all that can be said at the current time.
Regards
Andy
Very true Andy.For now, these are just conjectures.
anything in the pipeline about having 2x 3 core dies on one chip for a 6 core cpu? now that would be a good seller.
DFI P965-S/core 2 quad q6600@3.2ghz/4gb gskill ddr2 @ 800mhz cas 4/xfx gtx 260/ silverstone op650/thermaltake xaser 3 case/razer lachesis
I would argue the opposite, the monolithic design has forced them into concieving the tri-core... it is a yield manuever, if yields were hunky-dory I don't think we would have seen a tri-core announcement.
Yeah, Intel calls it a "yield improvement technique".
http://www.wired.com/techbiz/people/...urrentPage=all
Posting Permissions
Reply With Quote
Bookmarks