Effects of asynchronous FSB and Memory speed?

[cardart]

Hi everyone. I bought the dell inspiron B130 laptop with Pentium M 1.7 mhz. it is advertised at 533mhz FSB and 533 Memory speed.

I later found out that it is running at 533 fsb BUT only 400 mhz memory speed. Dell wouldnt do anything about it because the motherboard only supports 400 mhz memory speed.

anyways, my question is, how does this affect performance?

i've heard that a 400mhz fsb with 400mhz memory will run faster than a 533mhz coupled with only 400mhz memory speed because it is running synchronously rather than asynchronously. is this true?

my only option is to return the laptop alltogether, dell wouldn't fix the issue. Or i could just keep it if performance isnt that much affected by it.

thanks guys

 

[Hacp]

IIRC, Pentiums have a quad pumped FSB. Do the memory is actually doing more than the FSB limit IIRC. 133*4= 533. The memory will have to be below 133FSB or DDR266 in order for the front side bus to be affected negativly in performance.

 

[cubby1223]

You have to take the "fsb" numbers and revert them back to actual clock speeds to see what matches.

intel 800fsb=200mhz
intel 533fsb=133mhz
intel 400fsb=100mhz

ddr400=200mhz
ddr333=166mhz
ddr266=133mhz
ddr200=100mhz

And you can see that the match is actually with 533fsb & ddr266 ram.

 

[bob4432]

i have a ibm t42 with a p-m 1.6 400MHz cpu. ibm/lenova put pc2700 ram in it, which is fine, but i thought it was weird that insted of running it 1:1 and having faster timings, they run it 3:5 so even though the fsb is 100MHz, the memory is running @ 166MHz....weird

 

[cardart]

i dont understand what ur saying.

but the FSB is 133 x 4 which is 533 mhz.

and the memory speed is 100 x 4 which is 400.

it uses DDR2 200 pin memory rated at 533 mhz on the memory stick itself, but the motherboard only supports up to 400 mhz.

if this helps explains it better.

thanks for the reply

 

[Jiggz]

I read this before about the adverse effect of asynchronous FSB and Memory. But I think Intel fixed it with a "hub interpreter" which reduces the affect. Although the performance is still behind compare to a synchronous FSB and memory.

 

[cubby1223]

Well I'll be the first to admit I don't know much about ddr2 speeds, as it hasn't made it's way over to AMD's side of the table. But regardless, ddr2-400 runs just as fast as ddr400, doesn't it?

Originally posted by: bob4432
i have a ibm t42 with a p-m 1.6 400MHz cpu. ibm/lenova put pc2700 ram in it, which is fine, but i thought it was weird that insted of running it 1:1 and having faster timings, they run it 3:5 so even though the fsb is 100MHz, the memory is running @ 166MHz....weird

But I can answer this - benchmarks showed early on that cpus with a 100mhz or 133mhz clock benifited when the memory was run at higher speeds, at 133mhz or 166mhz. But for some reason, running the memory at asynchronysly at 200mhz hurt performance and realy slowed the development of ddr400.

So yeah, in some situations async is good, given that the memory is running faster than the cpu clock.

 

[bob4432]

Originally posted by: cubby1223
Well I'll be the first to admit I don't know much about ddr2 speeds, as it hasn't made it's way over to AMD's side of the table. But regardless, ddr2-400 runs just as fast as ddr400, doesn't it?

Originally posted by: bob4432
i have a ibm t42 with a p-m 1.6 400MHz cpu. ibm/lenova put pc2700 ram in it, which is fine, but i thought it was weird that insted of running it 1:1 and having faster timings, they run it 3:5 so even though the fsb is 100MHz, the memory is running @ 166MHz....weird

But I can answer this - benchmarks showed early on that cpus with a 100mhz or 133mhz clock benifited when the memory was run at higher speeds, at 133mhz or 166mhz. But for some reason, running the memory at asynchronysly at 200mhz hurt performance and realy slowed the development of ddr400.

So yeah, in some situations async is good, given that the memory is running faster than the cpu clock.

what type of increase in speed - 2% 0r 20%? my server machine is a xp2000-133MHz cpu, and i am running pc3200 @ 133MHz, with 2-2-2-5 timings but could easily bump the memory up to 166 and still have fast timings.

 

[cardart]

so no one knows about the effects specifically? any benchmarks or numbers for comparison? thank you

 

[Fox5]

The P-M's don't seem to have a large performance effect from async memory.

However, Athlon XPs do, don't run an AXP async. For that matter, the AXP FSB is only double pumped, so there's no reason to run the memory faster than the FSB.

 

[bwnv]

Originally posted by: Fox5
The P-M's don't seem to have a large performance effect from async memory.

However, Athlon XPs do, don't run an AXP async. For that matter, the AXP FSB is only double pumped, so there's no reason to run the memory faster than the FSB.

I think I answered this just yesterday? Anyway, to add to the above, A64's and AMDdual cores can run either way due to the on core memory controller. (just in case someone's looking here due to that question)

 

[cubby1223]

Originally posted by: bob4432
what type of increase in speed - 2% 0r 20%? my server machine is a xp2000-133MHz cpu, and i am running pc3200 @ 133MHz, with 2-2-2-5 timings but could easily bump the memory up to 166 and still have fast timings.

For a 266fsb cpu, ddr333 speed would be the best, and most motherboards set that up that speed by spd when you have pc3200 ram & a 266fsb cpu.

For benchmarks, you would really have to look in the archives here at Anandtech, or Tomshardware, or another site - ddr333 started coming out back in 2002 if I remember correctly - and look for when the KT333, KT400, & KT400A chipsets were first released, that's where the benchmarks would be.

 

[Fox5]

Originally posted by: cubby1223

Originally posted by: bob4432
what type of increase in speed - 2% 0r 20%? my server machine is a xp2000-133MHz cpu, and i am running pc3200 @ 133MHz, with 2-2-2-5 timings but could easily bump the memory up to 166 and still have fast timings.

For a 266fsb cpu, ddr333 speed would be the best, and most motherboards set that up that speed by spd when you have pc3200 ram & a 266fsb cpu.

For benchmarks, you would really have to look in the archives here at Anandtech, or Tomshardware, or another site - ddr333 started coming out back in 2002 if I remember correctly - and look for when the KT333, KT400, & KT400A chipsets were first released, that's where the benchmarks would be.

I think VIA based chipsets did better with the memory speed faster, but nvidia based chipsets did best with synchronous timings. (and thoroughly whooped via chipsets in performance)