Real world experience with Intel CPUs

A reader posted the following based on his many years of experience:

I live in the “Real World” not the “Benchmark World.” The problem I have with the Intel CPU is that it is real fast until it reaches “Critical Mass” where it suddenly stops. (Five minutes to open Outlook, I couldn’t connect with Terminal services either.) AMD, however, slows proportionally to the load, and when dealing with an Outlook memory leak putting the CPU at 100% usage recently, I was still able to remote in and fix the problem.

I fully agree. I had used Intel servers before, in a lot of situations, the server got overloaded, reaching 100% CPU utilisation, when that happened, it was impossible to login remotely. I had to call the datacenter and had the locked up Intel server rebooted. I never had similar problems with Opterons.

In one of Bob Colwell's stanford lectures, he admitted that the Intel CPU design uses an instruction scheduling algorithm that often enters a kind of loop which it takes a long long time to get out. When that happens, the Intel CPU would appear locked up.

Primitive stuff.

Athlon 64 x2 4400+ (65watt) for $69 free shipping

Good deal indeed. AMD is truly flooding the market.

AMD FAB30 is being converted to 65nm

K10 on time to frag Intel.

Intel does not innovate, it steals AMD's ideas: x86_64, dual core, performance per watt, spinning off flash division. The following are items Intel has not yet copied: IMC, HT, true quad core...

Intel should be ashamed of itself

OLPC project founder has some basic education for Intel the monopolist. The OLPC laptop is more advanced than any Intel Core 2 powered machine.

AMD's K10 will blow away any Intel quad core, says Randy Allen.

Intel's front bus architecture in plain English

Nathan Brookwood, founder of research firm Insight 64, said the following about Intel's Core 2 quad: "It is as if you wanted to talk to your neighbour in your apartment building, and the only way to do it was to mail them a letter."

Why is Intel QX6700 quad core a $125 CPU

Look at this picture. Michael Dell was excited, I heard, he is going to switch to near 100% AMD by 1Q08, just one quarter ahead of Intel's projected BK.

I think AMD's seemingly painful 1Q07 was Hector's plan. By taking a huge hit in 1Q07, AMD builds massive amount of inventory to flood Q2 -- those $59 X2 3600s just keep coming wave after wave. Then in Q3, K10 is out to push all of Intel CPUs down below $100 (except QX6700, which will be sold at $125). Intel's planned price cut in July is too little-- they only cut 50%, they need a 75% cut.

Intel's Penryn may see some clock speed increases, but those will be way short of narrowing the huge IPC gap. Also keep in mind, AMD is migrating to better SOI transistors in 65nm, which will result in 40% clockspeed boost.

Rahul thinks AMD made Intel stronger than ever. How ridiculously wrong. He can't see that Intel is at its last breath. Every sign tells us that Intel is doomed. The reason is simple. AMD has become a major player in server, desktop and mobile and has sufficient capacity. All AMD needs is one killer chip and Intel will be mortally wounded. This is because Intel is huge animal which needs a lot of food to survive, once AMD cuts Intel's revenue by 50%, Intel will die within two quarters, insufficient time for making any adjustments. The situation for Intel is unlike others. SUN always has this Sparc niche to enjoy its meal, so it can last a long time dying slow. Intel and AMD CPUs are software and hardware (except MB) compatible, and AMD has become a smarter choice.

Now, you ask, why should Intel die?

The reason is simple: Intel has been lagging behind AMD for five generations for four years, and there is no sign of catching up. Intel is behind on (a) virtualized 64 bit computing (b) IMC (c) direct IPC (ccHt) (d) direct Core-Core communication (e) virtualized HyperTransport I/O.

The Core 2 architecture somehow reduced the gap in 64 bit and Core-Core communication. But it was a kludge, the solution can't be extended to quad core. Intel's 64 bit architecture is incapable of IOMMU and now AMD made it to IOMMU 1.2.

The K10 will widen AMD's five generation architecture lead.

Rahul needs a philoshopy lesson

Rahul expressed his disblief on the Journal of Pervasive 64 bit Computing. I read news every day, and I fully understand Rahul's feelings. He is an AMD fan. He likes to see AMD win, though he pretends to be fair to Intel. But, as almost everyone else, his vision is way too short.

I simply think on a larger time scale. When you look at problems at larger scale, the fluctuations get smeared out.

AMD is indeed in an enivable position. 1Q07 was AMD's moment of awakening. As a result, it has tons of x2 inventory that it can flood the market like hell, and it has the K10 killer right behind that flood, and DELL, HP, Lenovo, Acer, SUN, IBM and everyone else (except a few corrupt Japanese vendors) will hype K10 like crazy. Intel's fastest chip will be slower than AMD's slowest.

Hector Ruiz described the future quite figuratively: AMD will launch a Tsunami of new products. What Intel will see will not be a huge wave caused by a Typhoon, but a giant body of water 500 feet high and with wavelength of 5 miles -- the drowning liquid just keeps coming and coming.

I re-iterate my projection that Intel BK by 2Q08.

Intel quad will be too expensive

Intel plans to cut quad core price to $266, but that is way too expensive. Once K10 is out, Intel CPUs should all be sold at below $125. We all know this: Intel quad is not really 4, is more like 2.5.

AMD HD 2900XT put Nvidia 8800 GTX to shame

Massive price/performance advantage , indeed.

K10 2.4GHZ runs at 1.1 v

Core 2 needs 1.3v. AMD's 65nm dual stress linear strained silicon on insulator with memorization and embedded Silicon Germanium (e-SiGe) (more buzzwords) is better than Intel's bulk vanilla dumb 45nm.

AMD is simply smarter.

Fun video

Pretty good. Bill Gates laughed out loud seeing this: these idiots, paying me billions of dollars.

My $1999 K10 setup

Tyan S2927 board ($280). K10 Barcelona 2P CPU ($895), buy one get one later. Still money left for 4GB registered DDR2 and one AMD 2900XT card.

No need to wait for AM2 quad.

K10 clock increase is 470 MHZ to be exact

INQ's reported 500MHZ hike was an approximation. The true number is 470MHZ (Of course, we know they are statistically equal.) Basically ,AMD's 65nm is doing 45nm speeds, the beauty of SOI+other exotic tricks.

This test show that AMD's 90nm X2 5600+ uses 20% less energy than Intel's E6400.

I heard Patty is polishing his resume and starts to call his old college friends. The problem for him is, there are only three x86 CPU makers and neither AMD nor VIA is interested.

Is Jack Thompson a typical American lawyer?

Someone mentioned Jack Thompson is being disbarred in the comments. I had no idea who Thompson is. But, look at what I found (see the docket at the end). Thompson practiced 30 years of law, but he acted like a total moron.

His excuse was that this was the first time he went federal... unbelievable. I thought in law school Civil Procedure is a required course.

Date Filed	#	Docket Text
03/30/2007	1	COMPLAINT against The Florida Bar Filing fee $ 350. Receipt#: 957378, filed by John B. Thompson.(ls) (Entered: 03/30/2007)
04/05/2007	2	ORDER REASSIGNING CASE. Case reassigned to Judge Paul C. Huck for all further proceedings. Judge James I. Cohn no longer assigned to case.Signed by Judge James I. Cohn on 4/2/2007 and by Judge Paul C. Huck on 4/5/2007.(ls) (Entered: 04/05/2007)
04/11/2007	3	AMENDED COMPLAINT Also Adding New Defendants against all defendants, filed by John B. Thompson.(Thompson, John) (Entered: 04/11/2007)
04/15/2007	4	AMENDED COMPLAINT (Second) against all defendants, filed by John B. Thompson.(Thompson, John) (Entered: 04/15/2007)
04/25/2007	5	AMENDED COMPLAINT (Third) Also Adding New Defendant against all parties, filed by John B. Thompson.(Thompson, John) (Entered: 04/25/2007)
04/26/2007	6	ORDER STRIKING 5 Amended Complaint filed by John B. Thompson,, 4 Amended Complaint filed by John B. Thompson, Signed by Judge Paul C. Huck on 4-26-07.(lc2) (Entered: 04/26/2007)
04/26/2007	7	Plaintiff's MOTION to Amend/Correct Complaint by John B. Thompson. Responses due by 5/10/2007 (Thompson, John) (Entered: 04/26/2007)
05/01/2007	8	ORDER denying 7 Motion to Amend/Correct. Signed by Judge Paul C. Huck on 5-1-07. (lc2) (Entered: 05/01/2007)
05/01/2007	9	Plaintiff's MOTION to Amend/Correct Complaint by John B. Thompson. Responses due by 5/15/2007 (Attachments: # 1 Exhibit State Petition for Writ of Mandamus# 2 Exhibit Second Amended Complaint)(Thompson, John) (Entered: 05/01/2007)
05/03/2007	10	Plaintiff's MOTION to Disqualify Judge Paul C. Huck by John B. Thompson. (Thompson, John) (Entered: 05/03/2007)

AMD quad to release at 2.8GHZ

Folks were dancing. There will be an AM2 quad at 2.9GHZ at launch.

The K10 will make all Intel CPUs sold in a narrow window of $65 to $125.

The following is a pricing list after K10 launch:

K10 quad (AM2) 2.9GHZ: $995
K10 quad 2.7GHZ: $795
K10 quad 2.5 GHZ: $595
K10 quad 2.3 GHZ: $395
K10 dual 2.9 GHZ: $695
K10 dual 2.7GHZ: $495
K10 dual 2.5 GHZ: $295
K10 dual 2.4GHZ: $195
Kentsfield 2.9GHZ: $125
Core 2 Duo 2.9GHZ: $105
X2 3GHZ: $101
Core 2 Duo 2.66GHZ: $110
X2 2GHZ: $65
Pentium D: $(-1)*

AMD FAB36 will be dedicated to K10 and Turion production. I expect dual core K10s be used for mobile. FAB30 will be dedicated to X2 production at 90nm, and it will become a 45nm FAB in 2Q08. Chartered should be converted to 65nm production soon.

Intel will pay $1 for each Pentium D sent to landfill. No one will buy a PC with Pentium D when Core 2 Duos will be dirt cheap. Remember, the CPU is just one component of a PC.

AMD may construct three more FABs

People may ask where the money is for these FABs. I tell you, that's never an issue. As long as AMD has the technology and OEM support, it can get 10 billion dollars to finish the FABs.

Don't forget that all of Intel's profits belong to AMD once Intel found guilty and is liable for treble damages.

Someone posted a benchmark showing Opteron faster than Woodcrest Xeon in CPU intensive SQL tests. But, this is only a test. In real life applications, consider yourself lucky if your Woodcrest Xeon did not crash randomly.

Intel double die solutions cause servers to crash

Intel multi-core is definitively not enterprise worthy, imagine a random circuit problem during a mission critical operation. The following servers are affected:

HP Proliant DL369 G5
HP Proliant DL380 G5
Dell PowerEdge 1950
Dell PowerEdge 2950

HP and DELL may be sued for product liability. Is Intel going to pay for the defenses?

AMD X2 6000+ frags Intel quad core Qx6700

In Matlab, Pro/Enginer, SolidWorks, and Apache web server benchmarks, the X2 6000+ frags the Intel QX6700 quad.

In file compression, audio encoding, the difference is around 5%. The only thing Intel qaud showed some real advantage is video encoding.

K10 will whip the crap out of Intel's double cheeseburger architecture.