• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Quick question about Celerons vs. Pentiums

K

Ken90630

Golden Member
I'm a bit embarrassed to be asking this, 'cuz I should know off the top of my head, but I never really paid much attention to Celerons so I guess it's time to ask and find out once and for all.

Briefly, what's the difference between Celerons and Pentiums, and in what applications/usage are Celerons noticeably slower?

I know Celerons have smaller caches, but how does that affect their speed? Does it relate to multi-tasking or ???

For discussion purposes, let's stick with Celeron D's and the latest Pentiums. Thanks in advance. 🙂

 
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
 
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

You can fit a Celeron D in a socket 3? :Q 😕
 
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

A FGA775 Celeron D (which uses DDR2) will only work on a FGA775 board. And the reason the smaller cache hurts performance is because if the information the CPU wants is not in the cache, it has to access the much slower DDR2 ram. (Any ram is slower than a processor's cache, not just DDR2.)
 
Originally posted by: Leper Messiah
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

You can fit a Celeron D in a socket 3? :Q 😕
Click to expand...


That's what I was wondering.
 
Originally posted by: CheesePoofs
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

A FGA775 Celeron D (which uses DDR2) will only work on a FGA775 board. And the reason the smaller cache hurts performance is because if the information the CPU wants is not in the cache, it has to access the much slower DDR2 ram. (Any ram is slower than a processor's cache, not just DDR2.)
Click to expand...

There are plenty of 775 boards that use DDR..I am using DDR with my prescott...
 
Originally posted by: stevty2889
Originally posted by: CheesePoofs
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

A FGA775 Celeron D (which uses DDR2) will only work on a FGA775 board. And the reason the smaller cache hurts performance is because if the information the CPU wants is not in the cache, it has to access the much slower DDR2 ram. (Any ram is slower than a processor's cache, not just DDR2.)
Click to expand...

There are plenty of 775 boards that use DDR..I am using DDR with my prescott...
Click to expand...

I never said there weren't. All I was tryign to say was a Celeron D will not work on any board, it will only work on a 775 board if its a 775 Celeron D and a 478 board if its a 478 Celeron D. It won't work on the same board a 486 would work on.

Acutally, I just reread his post and I think he meant a socket 478 board instead of a board for a 486 CPU.
 
Ya lol I didn't bother to google 478 and the first 4XX word I thought of was 486 🙂 . Also DDR2 memory running at 667 2-2-3-X is not Slow IMO... Way faster than similar DDR memory. The main hinderance to the Celeron IS the cache. Longer pipeline stages means that if the branch predictor predicts wrong, it has to start all over again, wasting alot of time. Caches generally help branch predictor misses because they provide way faster memory acess than ram does. Although faster ram will help, 1MB of cache will help more compared to 256 on the Celeron Ds.
 
celerons are designed for people who want a cheap computer. if intel made the celeron processor as good as the P4 Processor, everyone would get the celeron because of its price benefits.
 
Originally posted by: Hacp
Ya lol I didn't bother to google 478 and the first 4XX word I thought of was 486 🙂 . Also DDR2 memory running at 667 2-2-3-X is not Slow IMO... Way faster than similar DDR memory. The main hinderance to the Celeron IS the cache. Longer pipeline stages means that if the branch predictor predicts wrong, it has to start all over again, wasting alot of time. Caches generally help branch predictor misses because they provide way faster memory acess than ram does. Although faster ram will help, 1MB of cache will help more compared to 256 on the Celeron Ds.
Click to expand...

The problem with that is CAS2 DDR2667 RAM doesn't exist. The latency of DDR2 is higher than that of DDR. Most DDR2 is CAS4 or CAS3. Some is even CAS5. I can't find a single DDR2 module capable of running CAS2 at 533 MHz, let alone CAS2 at 667 Mhz.
 
Thanks, all, for the info and education. 🙂

One other thing: I know a little about how CPUs pre-fetch expected instructions. When a CPU does this (trying to anticipate what you're gonna ask the CPU to do next), does it store this info in the cache while waiting for you to ask for it? Or does it 'store' it somewhere else?
 
Originally posted by: Leper Messiah
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

You can fit a Celeron D in a socket 3? :Q 😕
Click to expand...

Yep, with a hammer......... Just kidding! 😛
 
Originally posted by: doc2345
Originally posted by: Leper Messiah
Originally posted by: Hacp
Celeron Ds are not limited to DDR2. They work fine on 486 boards. The main reason why celerons are so much slower are because of the lower FSB (533), less cache, which is a biggy because the Prescott has more pipelines, so it needs the cache to make up for wrong branch predictions, and finally because it does not have HT which is hyperthreading.
Click to expand...

You can fit a Celeron D in a socket 3? :Q 😕
Click to expand...

Yep, with a hammer......... Just kidding! 😛
Click to expand...

That was like the only way to get the 486's in. (Not really) But to get them out you had to almost pry them out, until the ZIF sockets came out... GOOD TIMES.

-Kevin
 
I remember that, used to use a flat blade screwdriver and slowly work the cpu out of the socket...... Oh the good ol' days.........
 
Originally posted by: doc2345
I remember that, used to use a flat blade screwdriver and slowly work the cpu out of the socket...... Oh the good ol' days.........
Click to expand...

Even then it was tought to get it out.

-Kevin
 
Originally posted by: Gamingphreek
Originally posted by: doc2345
I remember that, used to use a flat blade screwdriver and slowly work the cpu out of the socket...... Oh the good ol' days.........
Click to expand...

Even then it was tought to get it out.

-Kevin
Click to expand...



Yup i still remember and have my socket 3 mother board and it is in a box with a 486 dx still plugged into it. Man the awesome power of that CPU back in the day was great. But it was a pain to get out though because i had an SX in there prevsiously and i got mad at it and bent a bunch of pins trying to get the darn thing out. As said before "the good ol' days"

^_^
 
Pentium 4's used to be bandwidth limited until 800MHz came out. It needs more with dual-cores, but as far as single cores goes it doesn't.

Caches are tens of times faster in both LATENCY AND BANDWIDTH than RAM.
 
Uh... no. Even at 800mhz FSB, the Pentiums are still very bandwidth starved. Even the ones running at 1066FSB need more bandwidth. It is the nature of the architecture.

Dual Cores, are simply dying. When dual FSB comes out (or better yet Intels new chip) they will likely see fairly decent gains.

-Kevin
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top