Last message on previous page:

 
 
Yea, having a large cache is SOOOOOO not fair. I mean, AMD would never do that with their top line of chips, like the FX series. I mean, NEVER!
 
Cache is not a bad thing. The vast majority of benchmarks don't magically fit inside the cache. Cache results in real world performance gains. Fewer cache misses never hurt anyone. Complaining that Intel puts too much cache in their processors makes as much sense as claiming that AMD puts too many logic transistors in their processors.

intel is basicly going to kill amd.... then amd will basicly kill intel, the cycle repeats itself............. pentium3 compared to old old athlon processors

 
 
AM2 will be the desktop socket for a few years. DDR3 supposedly is pin compatible with DDR2. If you just got AM2 you are good to go for a few years. K8L WILL be on AM2 accordign to the Analyst Day article at Anandtech.com

Hey, don't forget, most of all, the huge cache.
That plays a major role in smoothing out the effect of the RAM clock and latency on Conroe's performance.

Unlikely.
Sure, K8L architecture is still largely unknown.
But for what it has been published, it looks like it will be competitive or outperform Conroe, but not by any huge margin.
So they better get it out quickly, before it's "too little, too late".

 
Yea, with that theory, its not worth getting anything more than a Celeron D, and that's what makes AMD's current advantage over Intel's bull****, am I right??

 
Hehe, they would if they could and if it would help them much... (wait until the die shrink)
 
k8 IMC help amd so it dosnt really need to have as much cache as intel -have- to have to perform good. Sure more cache would help but not as much as it helps intel. going over 1Mb l2 cache will help some but not much (comparied with the 1Mb one).
Fast ram helps AMD becouse the have an IMC (integrated memory controller) and small caches. When the cache isnt big enugh they go to the ram and then it needs to be fast.
Fast ram DOES NOT help intel that much for the same resons! Huuh?
Well intel have large caches (4MB) and is therefor able to keep most of what it needs in cache and as loong as the cache is big enugh then ram speed dosnt mather (that much).
 
And please all stop comparing pricing on an unrelised cpu (conroe) with cpus out now. Also count in that an Intel MB will cost much more then a camparable one for AMD. Amd will prob have to cut some prices esp at the lower end x2's.
 
Conroe is/will be a killer chip WHEN it comes.  
How much a performance hit will the 2Mb cache versions will take agains the 4Mb versions? Hard to say but ~5% or so in games is prob expected.
 
Then again if a 4Mb cache conroe is ~10-15% faster then A64
Then the 2Mb version is prob ~5-10% faster clock for clock.
 
Now thats faster, thats better and if u have it at the same cost (cpu/MB/mem) as what AMD can offer I will get one when I change next time.
 
But its not the killer it looked like when we read the first prev. of it.
AMD can and will compete (not beat) with current tech and might even beat it with faster chips after the dieshrink (christmas time).
But I would think that it isnt untill K8L then AMD will take lead again (GHz to GHz that is).
 
And sorry for the bad spelling/english

Unlikely.
Sure, K8L architecture is still largely unknown.
But for what it has been published, it looks like it will be competitive or outperform Conroe, but not by any huge margin.
So they better get it out quickly, before it's "too little, too late".
 
 
From what I've seen K8L will give 40% increase over K8. They're looking to increase IPC by 1-3 IPC. The new OoO bufer will give 10% along with the increased fetch. Look at how much better than Core Core 2 is.
 
Of course, it may not be 40% fatser than Conroe, but it will get at last 10-20% with the improvements to the core.

 
Link please.

 
Link please.
 
Word.

Unlikely.
Sure, K8L architecture is still largely unknown.
But for what it has been published, it looks like it will be competitive or outperform Conroe, but not by any huge margin.
So they better get it out quickly, before it's "too little, too late".
 
 
From what I've seen K8L will give 40% increase over K8. They're looking to increase IPC by 1-3 IPC. The new OoO bufer will give 10% along with the increased fetch. Look at how much better than Core Core 2 is.
 
Of course, it may not be 40% fatser than Conroe, but it will get at last 10-20% with the improvements to the core.
 
I think the key word here is hopefully get 40% improvement over K8.  It is so hard to tell what it will get based on numbers and specifications on a page somewhere.  As with Conore, it is best  to wait and see when they actually produce the chip and tests are run.  There is also another valid point that needs to be made here.  Everyone assumes that Intel will just ride on the current Conroe chip design and not make improvements for a few years.  Next year we should see chips based on the 45nm process from Intel.  AMD will only be at 65nm.  I also believe Intel is looking at a quad core processor next year as well.  I love AMD but don't count on Intel stopping with Conroe.  By the looks of things Intel appears to want the   market back from AMD and they may just get it.

my whole thing, is, that, for $500 you get a rock solid cpu. YES you will need a new mb, just as you would with a amd.
BUT $ for $, as we say, double the price, and YET slower, even in 1 test out of 10 test, STILL. DOUBLE the price..  
The FX is in a little trouble.  
And before anyone mentions dual FX64's.. WHATEVER, 1. THe MB would cost >300. 2. So you are looking for a >$2500 investment, for cpu and mb. That is not an option. And if it, then why not dual op's.. Which have new compitition from the 50xx series of xeon, which are based on the Dual Core..
 
So AMD has to put up, or stand by. lol   8O  8O  8O  8O [/quote]

 
No it won't.  Only in synthetic benchmarks using SSE.
 
Core has same OoO buffer, which most think is probably better than the K8L's.
 
Increased fetch won't do that much.
 
Increasing issue width from 3 to 4 as in Core brings 3-5% performance at best, increasing fetch alone won't be better, likely worse.
 
All it looks K8L adds to K8 in terms of microarchitecture is put what Core has in K8, like the OoO buffer and indirect branch prediction(which has been there since Pentium M, btw).

 
In synthetic apps.  But in synthetic apps Core already performs up to 4x better than K8 doesn't it??  And besides, Core already has what K8L will feature, and seems the implementation will probably be better according to current info.

 
 
Yes it does, lol synthetics for the 2.4GHz E6600 Core 2 Duo against the 2.8GHz FX-62 courtesy of FiringSquad.com:
 
Obviously this is just silly, but yeah some of the synthetic scores for Conroe are outrageous...
 

 

 

 

 
Yeah thats the market......... i still wonder if my quad opteron is worth it....as i can get it NOW.....hell my A64 3200 is fine by me... just want to multi-task like crazy....

Theoretical scores are important.  We all know they are not the end all and be all.  One thing to factor in is that MIPS is the number of instructions per second, however the more instructions in the instruction set architecture the slower the speed.  Alot of the numbers that are floating around I believe are flawed, particularly those of TomsHardware.  Here is a direct quote from their conclusion "...the Intel system had been preconfigured and didn't represent the final version".
 
It's going to be really interesting to see what the real numbers are going to be.

Good thing I've already 'ignored' you, I have no idea what you typed.

post of the day award