Nehalem IDF @ 3.2 GHz.

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,131
3,667
126
Originally posted by: Nemesis 1

What did I mis . What happened Yorkfield?????

not much unless you want to see a 4.2ghz 10hour prime stable clocked one. :T


But basically the low multi is making things very difficult for everyone. This is why that moral delima regarding the QX is no longer a delima. :T See first line if you dont believe me.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: bryanW1995
Originally posted by: zsdersw
Originally posted by: bryanW1995
sorry, that must have been you not aigo. Up to 200% is very nebulous, that's like me saying that we're giving "up to" 5000 off on wranglers. that could be 5000 off every single one or it could be $5000 off ONE wrangler and 0 off all the others. I just find it very difficult to see them getting ANYTHING REMOTELY approaching 100% efficiency on f@h with smt, but we'll see...

If the 200% figure were totally bogus, then why didn't they say 250%? Or 300%?

My guess is that the 200% figure happens in a very isolated and specific situation.. or, beyond that, *very* rarely.

I didn't say the number was bogus. In fact, I'm confident that it's legit, I'm also just as confident that it's not going to be perfect scaling for all apps. In fact, it stands to reason that an app like f@h that already taxes the shit out of a cpu will benefit the least from smt. however, I hope that I'm wrong b/c I'd love to have a 4 core nehalem with smt double up on a Q9450 @ the same frequency.

Well of course it's not going to be perfect scaling...this has been the crux of multithreaded programming since Gene Amdahl made is observation in the 60's...i.e. Amdahl's Law.

The point is that within the traditional confines of Amdahl's law, and later modification by Almasi and Gottlieb, you cannot get super-scaling...the best you can ever get is one-to-one scaling. So when Intel says they can get up-to a 200% for having 200% more threads they obviously mean to say that the SMT threads are every bit as good as the native threads, they are equivalent.

So for whatever Amdahl characterstics any given program has for scaling from 1-4 threads on a Yorkfield (be it perfect scaling, or 1/2 scaling or 1/10 scaling) the same scaling limits (not the same as scaling factor) will continue for threads 4-8.

Were this not true then they would have simply capped the upper end for perfect scaling at the hardware limited case...were SMT threads only work 10% that of a native thread (HT for example) then they would have said "up to 110%".

Saying up to 200% is obvious for the super course-grained highly scaling codes, such as povray or cinebench. But it works back from there too. Multi-tasking single-threaded apps is effectively the same thing as course-grained computing as well...so if you want to do 8 things at once then SMT stands to be very helpful. (I fall into this category, being a Beowulf cluster person)
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
I still not understanding I thought intel released cheaper Penryns. Ya got me confused here a bit. Do you exspect Intel to drop Merom prices to AMD X2 levels . Which would kill AMD. Whats Intel to do here . To me it looks like they are holding AMD head up out of the water so they don't drowned. 2 years ago AMD x2 was selling At price of over $300 dollars for the low end.

What are they selling for now. Who do you thank for this AMD or Intel . I say Intel . Amd comes out and Offers K10 black edition with unlocked mult. AMD guys were pissing there panties over the low cost an a Cpu that has little headroom and offered nothing in reality other than 4 cores on a Real life Qcore. We need AMD on the bottom to hold prices down . If AMD is on top instant price hike and Intel won't cut prices as they have know reason to worry as AMD can't only make so many cpu's . AND AMD would cry preditary pricing . and monoply abuse. If intel would have cut P4P prices to the bone like AMD did with X2 what would have happened . I think we all know the answer to that. You see AMD has an unfair pricing advantage because intel is a monoply. Intel can't cut prices to the bone or Hector will run crying to EU. Plus the present law suite would gain merit.

So intel offering rebates is bad but AMD selling CPU's below cost is OK. AMD is losing money . There second Fab is doing little but nothing .

I just don't get you guys. If intel does something its bad or useless . Kinda like the debates out in the forums now . About 2 cores Vs . 4 cores . Before K10 was released and found out to be the K9unit in disguise( Doggy bag required) A true Qcore was gods gift to man . Now that its found to be a doggy chip . These same people are saying we don't need 4 cores now. LOL!

Same thing with AMD 64 how long has it been out and how many apps. are supported. But it was hailed as a gift from heaven . LOL!

At least I am consistent. Not bowing down to the wind.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Would you really pay somebody to detract your competitors by calling them Nazi conspirators?
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
Alternatively, there might be a massively MT app which utterly pollutes the cache and dies with the current memory access scheme but gets 200% less constipated with lower latency access to RAM provided with the new on-die memory controller. That app would be a great showcase for the new CPUs, and I suspect it's a video encoder of some flavor.

15% single threaded IPC advantage over core2 *and* SMT that doesn't impact performance? I must admit it sounds too good to be true, but sign me up for at least one anyway.

As far as now time being grim for the value buyer -- say WHAT? $50 CPUs that breathe fire, $200 quad cores that perform as well as any commodity CPU and run stable on a $60 motherboard... Seriously, I've spent more on dinner. I don't think we'll see times this good any time soon. If you want grim, set the wayback machine to spring or summer of 2006.

 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Here is a fact . I can BS with the best of them . But out and out lieing not happening. When I was young I was mean lieing thief. Thing is I was really Good at it. But I not real sure what happened but I fell in love with my wife when I was like 14. She wouldn't have nothing to do with me. But I was consitanent in my early teens and finely one her over because of how I held here when dancing together. She laid down some rules . I followed them to the letter.

I haven't done anything really wrong since I was 17 . I pay for everthing even if its offered for free. I have gotten offers from hardware vendors but refused them.

I owe no one nothing I am in no ones pocket and never will be. I am to old now to change . I own everthing I have free and clear I have zero debt. Last year between the 2 of us we neted 6 figures and the year before that and before that all the way to 1990. I don't work now because of medical conditions my bad luck but I make more now investing than I ever made working I have had a good life. I am not happy about the world nor should I be. But my little world has been good. Human sufferring again is a matter of perspective .


So just because I am a believer that as far as hardware goes Cpu wise Intel is the smartest play for me. Why because I believe Intels the leader and has the resources to push harder and further than anyone else can . Its that simple. So intel gets my cpu and chipset dollars. Plus what I already stated about IBM/AMD.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
Originally posted by: CTho9305
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Would you really pay somebody to detract your competitors by calling them Nazi conspirators?



What ya don't like the trueth. What IBM and the HRCC did during ww2 was unforgivable.

IBM helped round up the jews and none christians during ww2. Its a FACT.

So if what Hitler did was so evil are you ready to so easily forgive him . Not a chance you say. So why ya giving IBM a free pass. Look what IBM did to Intel on x86 . They used Intel to put dec out of business. IBM is a ruthless evil company period. Strange how AMD fabs are in germany . What part oif Germany are these fads located. What's the political history of this area of Germany.

now back to the topic. If what Hexus says is true. 3.6 ghz on 4 core die Higher performance perclock than penryn at same clock by 15%-30% 4 cores 8 threads new hyperthread. Whats not to like maybe I hope Intel will give us a taste of its true power by doing some benchies tomorrow or today I should say today I guess. But its kinda early considering fall idf is just before Nehalems release. Now if we here nothing befor fall IDF could be the P4 team dropped the ball once again. But Intel seems pretty upbeat about Nehalem . But I do recall the prescott hype also.
 

DerwenArtos12

Diamond Member
Apr 7, 2003
4,278
0
0
Originally posted by: CTho9305
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Would you really pay somebody to detract your competitors by calling them Nazi conspirators?

Only if they couldn't trace it back to me :laugh:
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,131
3,667
126
Originally posted by: DerwenArtos12
Originally posted by: CTho9305
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Would you really pay somebody to detract your competitors by calling them Nazi conspirators?

Only if they couldn't trace it back to me :laugh:

Hasnt been a first time ive been hated by a Manufactorer. Im sure koolance, TT, SilentX, and innotek have me on there offical Ban list. :p

i'll talk mad crap about any company that lies or tries to push crap products. Thats just how objective i am.

and yes you can quote me when i bash a product. Because i would love the vendors to read it and try to come back with a response.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: Nemesis 1
Originally posted by: CTho9305
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Would you really pay somebody to detract your competitors by calling them Nazi conspirators?



What ya don't like the trueth. What IBM and the HRCC did during ww2 was unforgivable.

IBM helped round up the jews and none christians during ww2. Its a FACT.

So if what Hitler did was so evil are you ready to so easily forgive him . Not a chance you say. So why ya giving IBM a free pass. Look what IBM did to Intel on x86 . They used Intel to put dec out of business. IBM is a ruthless evil company period. Strange how AMD fabs are in germany . What part oif Germany are these fads located. What's the political history of this area of Germany.

now back to the topic. If what Hexus says is true. 3.6 ghz on 4 core die Higher performance perclock than penryn at same clock by 15%-30% 4 cores 8 threads new hyperthread. Whats not to like maybe I hope Intel will give us a taste of its true power by doing some benchies tomorrow or today I should say today I guess. But its kinda early considering fall idf is just before Nehalems release. Now if we here nothing befor fall IDF could be the P4 team dropped the ball once again. But Intel seems pretty upbeat about Nehalem . But I do recall the prescott hype also.

What I don't quite follow is why you hold a grudge against a company that doesn't have a single employee working in it that was involved in anything to do with WW2.

How is it even remotely relevant to demonize current-day IBM just because of something a handful of decision makers took it upon themselves to do in the 1940's?

Why would you penalize hard working good people at IBM these days by not spending money on their products? Surely not a single employee at IBM today worked there in the 1940's.
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
Originally posted by: Idontcare
Originally posted by: Nemesis 1
Originally posted by: CTho9305
Originally posted by: bryanW1995
nobody here is arguing that you EVER fail to promote intel. please tell me that at least give you free hardware...

Would you really pay somebody to detract your competitors by calling them Nazi conspirators?



What ya don't like the trueth. What IBM and the HRCC did during ww2 was unforgivable.

IBM helped round up the jews and none christians during ww2. Its a FACT.

So if what Hitler did was so evil are you ready to so easily forgive him . Not a chance you say. So why ya giving IBM a free pass. Look what IBM did to Intel on x86 . They used Intel to put dec out of business. IBM is a ruthless evil company period. Strange how AMD fabs are in germany . What part oif Germany are these fads located. What's the political history of this area of Germany.

now back to the topic. If what Hexus says is true. 3.6 ghz on 4 core die Higher performance perclock than penryn at same clock by 15%-30% 4 cores 8 threads new hyperthread. Whats not to like maybe I hope Intel will give us a taste of its true power by doing some benchies tomorrow or today I should say today I guess. But its kinda early considering fall idf is just before Nehalems release. Now if we here nothing befor fall IDF could be the P4 team dropped the ball once again. But Intel seems pretty upbeat about Nehalem . But I do recall the prescott hype also.

What I don't quite follow is why you hold a grudge against a company that doesn't have a single employee working in it that was involved in anything to do with WW2.

How is it even remotely relevant to demonize current-day IBM just because of something a handful of decision makers took it upon themselves to do in the 1940's?

Why would you penalize hard working good people at IBM these days by not spending money on their products? Surely not a single employee at IBM today worked there in the 1940's.

I don't know about that. Somebody who started just after high school might have joined at age 18 in 1944, personally convinced the CEO and board of directors to help Hitler, and chosen to never retire. There are lots of 82-year-olds still in industry. :roll:
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,787
136
tel is on record saying single-thread IPC is improved 15% but multi-threaded performance is improved "up to 200%", comparing Yorkfield vs Bloomfield.

Actually, Intel said 10-30% for single-thread and 20-100% for multi-thread performance improvement. Which if we average it out=10+30/2=17.5% Since Merom got 20% IPC increase over Yonah, we can assume Nehalem will be 20% faster over Merom.

I read (maybe from aigo) the other day that smt is supposed to be MUCH better than ht was. I used to run seti@home on an old p4 with ht and it only got about 10% more ppd than a normal p4 at 3.0. Supposedly smt is going to be a LOT better than that. I guess we'll see...

There is a reasonable evidence to believe that Nehalem would be MUCH better at SMT than Pentium 4 did(actually HT is Intel's term for SMT, but that's besides the point).

Pentium 4 had two major flaws that blocked SMT from showing true power.

1. Pentium 4 had 1 decoder, ONE decoder!! The Trace Cache was made to make up for it, but then the cache hit rate was said to be low(50-60%). What happens when cache hit rate is low?? The CPU is effectively 1-issue machine lots of times. But 1-issue isn't at the point where there is ILP bottleneck. Therefore, if the SMT thread was running, sometimes it was trying to take advantage of the sometimes-1-issue CPU, so the benefit of SMT was greatly diminished
2. Pentium 4 had a thing called Replay. Basically because the pipeline was so long Intel had to find a way to speculatively process an instruction. Replay was a feature to basically "replay" the execution potentially saving clock cycles and effectively reducing the pipeline stages... when it WORKED!!!

But then when it didn't work(like a cache miss), it would take valuable processing cycles processing that instruction and the CPU would be utilized doing basically nothing during that time. Sometimes, the replay system would "replay" for numerous amount of times!!

Because the CPU was filled up with processing an instruction already, getting another thread meant that the original thread lost performance. So that's why there was sometimes a drastic performance loss and sometimes pretty good performance for SMT.

Nehalem has NONE of the above mentioned problems. Plus, Nehalem is wider, has substantially more memory bandwidth.

That's why even Intel's claims are much much better for SMT gains for Nehalem than it was for Pentium 4. Pentium 4 was claimed to get up to 30% increase with HT. Nehalem, Intel says will get 20-100% increase, but if we disregard the IPC increase the actual improvement will be something like 0(in single thread)-50%.

However we look at it, there is something amazing coming on the horizon :D.