Intel Plans 3.40 GHz Celeron E3900 for 2010

[techfan]

I just wish they would stop calling them celerons, makes it so unsellable.
look how fast my rig is, its amazing..... yeah sure oh but its got a celeron inside NO THANKS
Just call it by its E number

 

[Phædrus]

dont worry about it. (cause he's wrong)

Phædrus: show me anything - anything, with reliable testing that has a speed difference, with the only change being the multiplier. same CPU frequency, same ram frequency, same HT link if on AMD.

I've seen people prove time and again that cache makes a difference (if not in every application) but in all the years i've been using PC's - back when multipliers were first introduced i have never, EVER seen ANY evidence that higher multis cause poor performance.

I'm not *saying* it gives poor performance. I'm saying the performance improvement per increase in multiplier is logarithmic. Going from a 1:1 bus-CPU ratio to a 1:2 ratio is a great improvement in performance. Going from x12 to x13 is a minimal improvement. Going from x12 to x17 is a moderate improvement at best. This processor will probably have really good performance. But because of the smaller cache and low bus rate compared to the CPU speed this CPU may only perform about as well as an E7500 (guesstimate! not tested!), even though it's clocked almost .5GHz higher.

The introduction of running the CPU at a ratio instead of on par with the bus was a great improvement in CPU design. It's a lot easier to design a CPU with a higher multiplier than it is to make one with a faster bus rate. And up to x8-x12 it gives good increases in performance. But when you go past that the curve gets shallower, it levels out some, it still increases as the multi goes up but not nearly as fast as it did for lower multipliers.

@JrRacinFan: I understand, and we're both pretty much arguing the same side of the argument. But threads like this (and the thread at TSF that I linked this from) get read by a lot of people over time. You google "Intel E3900" and this is one of the first results. So I'm mostly trying to put information out there for those who don't know a ton, who maybe just know the basics.

 

[newtekie1]

We'll I'll be buying one, so I'll test your theory Phædrus, but I don't believe you are entirely correct.

Obviously the higher bus speed gives better performance, so two processors clocked at the same speed can perform differently, the one with the lower multipler and higher bus speed will perform better, this is common overclocking knowledge. If you max out the chip before the board, lower the multipler to push a little more performance out of it. It is for this very reason that I lowered the multipler on my Q6600 to 8, and ran the board at 400MHz, the chip would only do 3.2GHz. The same is true of my E1400.

However, I don't believe the issue with the high multipler is as big as you make it seem. And overclockers will definitely be lowering the multiplier on this thing to get higher FSB speeds. I mean just to get to 266FSB, the processor will need to run at 4.5GHz, and I highly doubt this processor will be good for that speed on reasonable cooling(though maybe).

Essentially what Intel has done is given us the closes thing we are going to get to a cheap unlocked Wolfdale. Having a multipler that can vary anywhere between 17 and 6 is going to be fun.

 

[Mussels]

We'll I'll be buying one, so I'll test your theory Phædrus, but I don't believe you are entirely correct.

Obviously the higher bus speed gives better performance, so two processors clocked at the same speed can perform differently, the one with the lower multipler and higher bus speed will perform better, this is common overclocking knowledge. If you max out the chip before the board, lower the multipler to push a little more performance out of it. It is for this very reason that I lowered the multipler on my Q6600 to 8, and ran the board at 400MHz, the chip would only do 3.2GHz. The same is true of my E1400.

However, I don't believe the issue with the high multipler is as big as you make it seem. And overclockers will definitely be lowering the multiplier on this thing to get higher FSB speeds. I mean just to get to 266FSB, the processor will need to run at 4.5GHz, and I highly doubt this processor will be good for that speed on reasonable cooling(though maybe).

Essentially what Intel has done is given us the closes thing we are going to get to a cheap unlocked Wolfdale. Having a multipler that can vary anywhere between 17 and 6 is going to be fun.

the reason they perform higher at the higher bus speed is usually due to the ram being clocked higher with it, or because the higher clock is at a 1:1 divider - theres less delay between CPU and memory.

 

[newtekie1]

the reason they perform higher at the higher bus speed is usually due to the ram being clocked higher with it, or because the higher clock is at a 1:1 divider - theres less delay between CPU and memory.

And there is simply more bandwidth for the CPU to communicate with the chipset and everything else in the computer(you're right that it is mainly memory).

Either way, we agree that the higher bus improves performance, not because there is a dely caused by the multiplier, but because there is more bandwidth to communicate with the rest of the system.

 

[OnBoard]

Obviously the higher bus speed gives better performance, so two processors clocked at the same speed can perform differently, It is for this very reason that I lowered the multipler on my Q6600 to 8, and ran the board at 400MHz, the chip would only do 3.2GHz. The same is true of my E1400.

the reason they perform higher at the higher bus speed is usually due to the ram being clocked higher with it, or because the higher clock is at a 1:1 divider - theres less delay between CPU and memory.

And there is simply more bandwidth for the CPU to communicate with the chipset and everything else in the computer(you're right that it is mainly memory).

Either way, we agree that the higher bus improves performance, not because there is a dely caused by the multiplier, but because there is more bandwidth to communicate with the rest of the system.

I did the same with my E7200 and now with this E8400, both running with 400FSB. With E7200 the gain from going 266->333 was already big. You just need to start windows and the performance is so much greater. Going from OC to stock clocks and buss with hot summer everything felt sooo slow.

Bandwidth good, multi "bad"