RAM Myth?
Hey all, A couple of years ago when I was building a 440BX system, a friend on mine who's very knowledgeable in such things told me that I could go up to 384MB of PC100 SDRAM and still keep 8ns access speed.
Hey all,
A couple of years ago when I was building a 440BX system, a friend on mine who's very knowledgeable in such things told me that I could go up to 384MB of PC100 SDRAM and still keep 8ns access speed. But if I went over 384MB, the access speed drops to 60ns, even if they are all 8ns chips. The reasoning being that with that much RAM, the system takes a little longer accessing the memory.
Now was my friend incorrect? I've been trying to research this on the web to see if it's true or not, but I haven't discovered anything. The reason I'm asking is that I've got a chance to bump up to 512MB of RAM with my Dual 440BX board, but I don't want to do it if it means I take a speed hit on the access speed.
Thank you.
A couple of years ago when I was building a 440BX system, a friend on mine who's very knowledgeable in such things told me that I could go up to 384MB of PC100 SDRAM and still keep 8ns access speed. But if I went over 384MB, the access speed drops to 60ns, even if they are all 8ns chips. The reasoning being that with that much RAM, the system takes a little longer accessing the memory.
Now was my friend incorrect? I've been trying to research this on the web to see if it's true or not, but I haven't discovered anything. The reason I'm asking is that I've got a chance to bump up to 512MB of RAM with my Dual 440BX board, but I don't want to do it if it means I take a speed hit on the access speed.
Thank you.
Participate on our website and join the conversation
This topic is archived. New comments cannot be posted and votes cannot be cast.
Responses to this topic
I'm sure that's not right, because it wouldn't be able to keep up with the 100MHz bus then. At most, 60ns memory could run at about 16.67MHz. End result - the memory subsystem wouldn't work AT ALL if the memory ran at 60ns. Besides, the only effect that too much memory could have on the system would be slightly slower memory addressing, which has nothing to do with the memory latency. But I don't think that's even the case, since PC's use 32-bit memory addressing, which can address (potentially) up to about 4GB.
Windows 2000 Pro will use whatever memory is available at its proper speed.
Hardware and software limitations may have been what your friend referred to. What follows is not technically precise or certain but what I gleaned on this matter.
Motherboard limitations determined if 256 or 384 or 512 Megs of ram would be cached. This involved a serious slowdown when increasing ram over this limit because ram being read from the top implied uncached ram would be used first. This slowdown nust be what your friend referred to.
I was told by a knowledgeable computer technician that Windows 98 could not use effectively more than 256 Megs of Ram. Why would this be I wonder.
Hardware and software limitations may have been what your friend referred to. What follows is not technically precise or certain but what I gleaned on this matter.
Motherboard limitations determined if 256 or 384 or 512 Megs of ram would be cached. This involved a serious slowdown when increasing ram over this limit because ram being read from the top implied uncached ram would be used first. This slowdown nust be what your friend referred to.
I was told by a knowledgeable computer technician that Windows 98 could not use effectively more than 256 Megs of Ram. Why would this be I wonder.
Win95 couldn't cache up more than 64MB of memory. It would "use" 128MB of RAM, but it wouldn't address it as fast as 64MB. As far as Win98 goes, I did not see any performance increase when I went from 128 to 256MB of RAM. I would just attribute that to a weak caching system.
Win2K and NT will use your RAM at whatever speed is set in the BIOS of the mother board. And as far as that 60ns issue, I have never heard of it either. Since I have 640MB in a BX based server of mine, I have never seen a drop in performance.
------------------
Regards,
clutch
Win2K and NT will use your RAM at whatever speed is set in the BIOS of the mother board. And as far as that 60ns issue, I have never heard of it either. Since I have 640MB in a BX based server of mine, I have never seen a drop in performance.
------------------
Regards,
clutch
I notice the jump in speed from 128MB to 256 MB all the time in win98, just try to enable the machine role as "server" it does only means to let win98 allocate more ram for the HDD caching, and then you could make IE to browse in a separate process for each instance it runs, it would use up much more ram but as 256MB are still enough to cater for it, and stability of system goes up too.
I notice speed up in Photoshops, CAD programs, UT, NOLF, etc, with 128 to 256MB change.
I notice speed up in Photoshops, CAD programs, UT, NOLF, etc, with 128 to 256MB change.
Your friend is incorrect. The only problem you may run into is if the BX board cant use anymore than 384MB, but that was on an Intel 440BX board I used to have, so things could be different.
Increasing the ram should make a difference immeadiately, unless its a didly squat amount like 32 megs.
I agree with what the other guys said about this, except for clutch and tylau comments, but thats because I dont know anything about that. 95 is way back, man. CUViper's answer is very logical.
Increasing the ram should make a difference immeadiately, unless its a didly squat amount like 32 megs.
I agree with what the other guys said about this, except for clutch and tylau comments, but thats because I dont know anything about that. 95 is way back, man. CUViper's answer is very logical.
60ns thingy is way back into 430HX age where FP/EDO ram stuffing are stil reign, those 60ns thingy cant keep up with the 66MHZ FSB/memory bus at their date, so something called "memory interleaocking" came into play which make it possible in theory to incorporate memory as slow as 40ns into a 100MHZ memory bus....
CUViper is right. Everything on a computer runs on clock cycles. The different Clock cycles speed determine the frequencies. There is no way ram can adjust the output cycle speed. The clock pulses are always going to be there no matter how big the memory is. If the fsb is clocked at 100MHz then thats what it stays at. The clock pulses will tell the cpu to write to the ram. There may be a delay, but timing will always be on the money.
[This message has been edited by CrazyKillerMan (edited 14 March 2001).]
[This message has been edited by CrazyKillerMan (edited 14 March 2001).]
