What makes Intel CPU faster ?

Mir96TA

Golden Member
Oct 21, 2002
1,950
37
91
I mean I have tested Core2 Dual Core with LinX and with excat same Core and clock speed I did AMD Phenom II
Even AMD Phenom II have more Cache, Faster memory and newer processor (Release Date wise) it is slower compare to Intel.
I am wondering what makes Intel so much faster ?
Here is what I have tested
Lynx1055at31Gigwith6Core.jpg

Mind it cool and quiet brings the IDLE speed down.
It was running at 3.1 Gig
 

Farmer

Diamond Member
Dec 23, 2003
3,334
2
81
I'm really curious myself also! I don't really know much beyond a basic understanding of CPU architecture.
 

Martimus

Diamond Member
Apr 24, 2007
4,490
157
106
I see from the picture that the AMD processor you are running is a X6 1055, but you don't say what the Intel Processor is. Without knowing that, it will be dificult to answer your question.

Is LinX the only benchmark that you need information on? I'll try to look up what kind of code it is, and why it runs better on Intel processors. I would guess it has a SSE instuction that AMD doesn't have in the 1055, but to be honest I don't know anything about the program at the moment.
 

TuxDave

Lifer
Oct 8, 2002
10,571
3
71
These examples are mostly because I'm ALU centric.

Sometimes it's obvious, like AMD's BD FMA where a single instruction will be able to do a*b+c and Intel will need to do a*b followed by +c. Very similiar lines of thinking like AES or STTNI where it's new hardware support for a specific type of computation.

Sometimes it's less obvious, where it's more like removing penalties. How quickly can the result of one instruction feed into another instruction.

In a nutshell, tons of different ideas are thrown around, evaluated if it saves power or improves performance and subset makes it in. The performance gain you see in the end is the result of hundreds of ideas working together.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,787
136
LinX is a user-friendly interface for Intel's MKL Linpack library. Linpack is an extremely FPU-intensive program. You can almost say its a synthetic benchmark, since the results can be almost predicted by calculating before running it.

Properly optimized, it can achieve GFlops numbers(yea that's what it measures) 70-80% of theoretical max.

For modern Intel and AMD CPUs, the formula is:

Number of cores x Frequency in cycles x number of DP FP operations/cycle

For example, the Phenom II X6 clocked at 3GHz:

6 cores x 3GHz x 4 DP FP operations/cycle = 72 GFlops

You are getting 60GFlops. That's a figure Core 2 Duo shouldn't be able to reach unless its running at 8+GHz.
 

Mir96TA

Golden Member
Oct 21, 2002
1,950
37
91
Sorry I was little late to do a second Post
Here is a Intel CPU E3110 (E8400) running at 3.1 Gig
Here is a Pic
Lynx.jpg
 

Mir96TA

Golden Member
Oct 21, 2002
1,950
37
91
Please dis reagard Orig Pic of AMD. I just wants to see How fast 1055T will run at 3.1 Gig with 6 Core. (I wanted to see scaliblity) with same speed and adding 4 cores does not match
Any rate here is the Pic of AMD running at 2 Core
Lynx1055at31Gig.jpg
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,067
3,574
126
Overclocking made so easy, even your grandma can do it:
http://www.youtube.com/watch?v=_GR5_X1CfUA

And at first we honestly thought it was a joke.. but:
http://forums.anandtech.com/showthread.php?t=2147593

That guy accidentally ate a whole GHZ.. lol...


But seriously.. its the arch, and the way programs were written for the arch.
There are some programs which an AMD will spank an intel, however those programs arent very common in what every day people use.

The every day software, is just better ran on the intel arch... like encoding... but the SMP stuff, is laid down on the AMD side still.

Also when looking at budget... AMD always has an advantage... but raw speed, with the ignorance of budget... intel has the floor.
 
Last edited:

Mir96TA

Golden Member
Oct 21, 2002
1,950
37
91
I mean they both running at 64 Bit OS and Application
They both have a same Class Codes well AMD have 3DNow. I am not sure how much more it helps over what it is already have.
Specs wise I see AMD processor have more edge like more Cache, Faster Ram, New Processor and Newest Chipset. I was thinking Prob AMD creeping faster then Intel.
Intel have Older CPU, Less of a Cache, Missing a Code (No 3D Now) Old Chipset, and slower Memory.
BTW OS is Win7 Pro 64Bit Sp1
 

Mir96TA

Golden Member
Oct 21, 2002
1,950
37
91
BTW fellows I am not looking a scablity of O.C.
That is tottaly different ball game
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,067
3,574
126
BTW fellows I am not looking a scablity of O.C.
That is tottaly different ball game

so your looking at raw to raw clock performance?


You really cant do that because intel cheats with something called turbo on, which AMD does not have.
 

Mir96TA

Golden Member
Oct 21, 2002
1,950
37
91
so your looking at raw to raw clock performance?


You really cant do that because intel cheats with something called turbo on, which AMD does not have.

I wanted to see what making Raw Speed so much faster on Intel
My Intel Didn't had Turbo
My AMD did had Turbo which I disable it. along other 4 Cores
 

podspi

Golden Member
Jan 11, 2011
1,982
102
106
Err... the PhII is a lot faster in that "benchmark" ...

Higher numbers are better :-D
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,067
3,574
126
C2D doesn't turbo.

Looks like he has C&Q enabled but not speedstep.

OMG u get a bonk aigo for free card for me missing the C2D part.

I wanted to see what making Raw Speed so much faster on Intel
My Intel Didn't had Turbo
My AMD did had Turbo which I disable it. along other 4 Cores

you know i honestly think a X6 could keep up with a C2D.

I have no clue..