flat assembler
Message board for the users of flat assembler.

Index > Heap > Which CPU is better?


Which CPU is better?
The Pentium 4
83%
 83%  [ 5 ]
The Celeron D
16%
 16%  [ 1 ]
Total Votes : 6

Author
Thread Post new topic Reply to topic
windwakr



Joined: 30 Jun 2004
Posts: 827
Location: Michigan, USA
windwakr
Which CPU looks better to you? The P4 is hyperthreaded, The Celeron runs 500 some Mhz faster than the P4. I Can't decide which is better. Although, it doesn't really matter. It's not like I'd be able to switch them around, I'd just like to know if my new main computer has a better or worse CPU(the celeron is in my new PC).


Btw: This new computer I bought was really cheap, $200 for a complete system. Nice Dell CRT monitor, speakers, mouse, nice keyboard. The computer has the Celeron in it, 512MB of memory(just ordered two 1GB sticks for it at about $15 each), A nice DVD burner with lightscribe, it has onboard graphics but I just ordered a cheap nvidia 8600 gts for $50($30 after rebate.)


EDIT: What is the EM64T on the Celeron? Is it 64bit?


Description:
Filesize: 48.32 KB
Viewed: 2745 Time(s)

my_cpus.png



_________________
----> * <---- My star, won HERE
Post 29 Jun 2009, 21:37
View user's profile Send private message Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17352
Location: In your JS exploiting you and your system
revolution
It depends. What do you want to do with it? One has more cache (and hyper-threading) so in some circumstances this will be "better" for a task. The other has more CPU MHz (but a slower memory bus) so in some circumstances this will be "better" for a task. But tasks also change so sometimes one will be "better" and other times the other will be "better". It depends.

There is no universal "better", so your question is not well formulated. I suggest the "best" way to find which is "better" is to use it for the things you want, and see which one performs the tasks you do in ways that you find "better".
Post 30 Jun 2009, 00:48
View user's profile Send private message Visit poster's website Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 2917
Location: [RSP+8*5]
bitRAKE
In general the P4 is better: faster bus, more cache.

The Celeron D has 64-bit capability, but how to keep the beast fed? True, at the cache level it is faster. IIRC, it also uses less power (65nm verses 90nm).

Can you share the cache latency numbers please? (Try System Information Viewer if you don't have some other similar tool.)

For reference here is my data:
L1 latency, 4 cycles
L2 latency, 15 cycles
main memory latency, 237 cycles!

In contrast my old Athlon (Barton) had only a 3 cycle latency for L1, and 11 cycles for L2, but it was a space heater.


Description: I got two of these bad boys under the same hood!
(Really 2.333Ghz but throttled down, ATM.)
Filesize: 24.19 KB
Viewed: 2703 Time(s)

CPU-Z.PNG



_________________
¯\(°_o)/¯ unlicense.org


Last edited by bitRAKE on 30 Jun 2009, 03:19; edited 1 time in total
Post 30 Jun 2009, 02:58
View user's profile Send private message Visit poster's website Reply with quote
windwakr



Joined: 30 Jun 2004
Posts: 827
Location: Michigan, USA
windwakr
Nice processor.

So, where in system information viewer can I find that info?
Post 30 Jun 2009, 03:17
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 2917
Location: [RSP+8*5]
bitRAKE
When you first run it, there should be a button on the right-hand side "Cache-0 Latency". It'll open a window and run a stride test - with the results concluded on the bottom.
Post 30 Jun 2009, 03:23
View user's profile Send private message Visit poster's website Reply with quote
windwakr



Joined: 30 Jun 2004
Posts: 827
Location: Michigan, USA
windwakr
Ok

P4:
L1 latency 4 cycles
L2 latency 28 cycles


Celeron:
L1 latency 4 cycles
L2 latency 28 cycles

They're the same.
Post 30 Jun 2009, 03:25
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 2917
Location: [RSP+8*5]
bitRAKE
Which means the Celeron D would truely be faster on cached data. The memory latency is in the lower right-hand corner of the table. I'd be interesting to compare the two.

There is also a latency tool on the CPU-Z web site:
http://www.cpuid.com/download/latency.zip
Post 30 Jun 2009, 03:38
View user's profile Send private message Visit poster's website Reply with quote
windwakr



Joined: 30 Jun 2004
Posts: 827
Location: Michigan, USA
windwakr
Where?

EDIT: heres output of the program you linked when run from the celeron, can't get anything more out of the p4 tonight, I don't want to start it back up.

EDIT2: ok, removing attchment.
Code:

Cache latency computation, ver 1.0
www.cpuid.com

Computing ...


stride  4       8       16      32      64      128     256     512
size (Kb)
1       4       4       4       4       4       4       4       4
2       4       4       4       4       4       4       4       4
4       4       4       4       4       4       4       4       4
8       4       4       4       4       4       4       4       4
16      4       4       4       4       5       5       5       4
32      4       5       9       16      28      28      28      28
64      4       5       9       16      28      29      28      28
128     4       5       9       16      28      28      28      28
256     4       5       9       16      29      30      30      31
512     4       5       10      18      33      52      73      75
1024    4       9       17      36      70      210     520     519
2048    4       9       18      35      70      210     518     521
4096    4       9       18      36      69      210     516     519
8192    5       9       18      35      71      213     520     518
16384   4       9       18      35      70      213     525     524
32768   5       8       18      35      70      212     528     519

3 cache levels detected
Level 1         size = 16Kb     latency = 4 cycles
Level 2         size = 256Kb    latency = 28 cycles
Level 3         size = 512Kb    latency = 66 cycles
    

_________________
----> * <---- My star, won HERE


Last edited by windwakr on 30 Jun 2009, 16:23; edited 4 times in total
Post 30 Jun 2009, 03:41
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 2917
Location: [RSP+8*5]
bitRAKE
The 520 in the lower right of the table. The CPU-Z tool seems flaky to me. Yet, this stride test isn't an exact measurement. The P4 should have better main memory latency because the bus is significantly faster. It'll be interesting to see if that is indeed the case.
Post 30 Jun 2009, 03:42
View user's profile Send private message Visit poster's website Reply with quote
windwakr



Joined: 30 Jun 2004
Posts: 827
Location: Michigan, USA
windwakr
So what exactly is the memory latency?
What exactly does it do? Whats the size and stride, I don't understand what this test is doing, and it doesn't have documentation.
P4:
Code:

Cache latency computation, ver 1.0
www.cpuid.com

Computing ...


stride  4       8       16      32      64      128     256     512
size (Kb)
1       5       4       5       4       6       4       4       4
2       4       5       4       4       6       5       4       4
4       4       4       4       4       4       5       4       4
8       4       4       4       4       4       4       4       4
16      4       4       5       4       5       5       6       4
32      6       8       11      20      28      28      28      28
64      6       8       11      21      24      26      28      28
128     6       8       11      19      26      28      28      28
256     6       8       12      20      29      29      30      30
512     6       8       11      21      29      30      31      33
1024    6       8       12      25      31      37      44      46
2048    6       8       16      30      63      141     308     315
4096    6       8       16      32      64      151     307     313
8192    6       11      16      32      64      185     309     312
16384   6       8       16      32      64      143     311     320
32768   6       8       16      32      64      144     313     319

2 cache levels detected
Level 1         size = 16Kb     latency = 4 cycles
Level 2         size = 1024Kb   latency = 31 cycles
    


Last edited by windwakr on 30 Jun 2009, 16:23; edited 1 time in total
Post 30 Jun 2009, 14:44
View user's profile Send private message Reply with quote
bitRAKE



Joined: 21 Jul 2003
Posts: 2917
Location: [RSP+8*5]
bitRAKE
Latency is the time it takes to complete an operation. For example, if we code MOV RAX,[hWnd]; the processor first checks the L1 cache for the value of hWnd. If the value is not present then it goes on to L2 cache, and finally main memory. A good analogy is how we keep stuff in our pants pocket, backpack, shelf - based on how frequently the item is needed.

The P4 would be able to stream data into the processor faster 2.8Ghz/319 = 8.7MT/s. Verses 3.3Ghz/520 = 6.4MT/s for the Celeron D. Furthermore the L2 cache is larger on the P4.

(I haven't looked at the source of a stride test, but I kind of have an idea how they work. Read A bytes, skip B bytes, loop until C bytes have been read. Might want to use code tags to ease reading of the table.)

Intel produces the Celeron D to create a market sector by both limiting the cache and bus -- it's feature compatible but slower than other processors of the same technology.
Post 30 Jun 2009, 16:14
View user's profile Send private message Visit poster's website Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  


< Last Thread | Next Thread >
Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum


Copyright © 1999-2020, Tomasz Grysztar. Also on YouTube, Twitter.

Website powered by rwasa.