- Jun 21, 2000
- 596
- 0
- 0
There seems to be conflicting info out there about this nForce Twinbank/Superstability controversy.
GamePC's article, nForce Superstability - Feature or Bug Fix?, is the only article I've seen that attempts to explain the memory hit the MSI K7N 420D Pro takes when Slot 2 is used.
But there is a contradiction in the article:
At the beginning of the article he states:
<< When straying from nVidia's memory compatability and configuration listings, problems may start to manifest themselves. Keep in mind, if you follow these lists, the problems that occur on the following pages do not apply to you. nVidia has made it quite clear to us, that this issue only occurs when you stray from the specified memory lists available from MSI and nVidia. >>
Then when you get down to the tests he states:
<< For the memory, we used standard, run of the mill Crucial/Micron PC-2100 DDR SDRAM (Non-ECC, Non-Registered). >>
Well that memory module is on nVidia's compatibility list so why is this testbed experiencing the problem?
Then he says:
<< Turns out, the problem is that when you use dual-sided memory the second slot (slot B) of the second DDR memory bank of any given nForce motherboard, and you will consistently see a 20-25% drop in memory performance. >>
But nowhere does he mention how he came to this conclusion or why that would be a problem - just flatly blames double-sided ram.
nVidia's technical briefs on the nForce's technology and TwinBank implementation doesn't mention anything about Superstability which I find suspect.:disgust:
NSF4 mentioned Overclockers.com's review of MSI's K7N420 Pro as another experiencing this memory issue. He's using 2 identical 256MB sticks of NANYA-chipped Apacer PC2100 CAS2 for testing. So he has "strayed" from the list, so according to GamePC, that explains (albeit in another blackbox kind of way) why he's getting the performance hits.
I've got 3 sticks of Crucial PC2100 DDR SDRAM that I got for $30 each a couple of months ago. They are listed on nVidia's compatibility list. That hurdles one stumbling block: 1) not activating SS and lowering the FSB rate. The only setback I can see with "compatible" memory modules is the other issue I haven't talked about yet: 2) the twinBank arrangement limits how much memory is transferred at 128 bit and how much by 64 bit. The bank with the least amount of total memory dictates how much from each channel will be 128bit and the rest at 64bit.
So I'll have 2x256 delivering at 128bit and 256 at 64bit and all at 266MHz.
GamePC's article, nForce Superstability - Feature or Bug Fix?, is the only article I've seen that attempts to explain the memory hit the MSI K7N 420D Pro takes when Slot 2 is used.
But there is a contradiction in the article:
At the beginning of the article he states:
<< When straying from nVidia's memory compatability and configuration listings, problems may start to manifest themselves. Keep in mind, if you follow these lists, the problems that occur on the following pages do not apply to you. nVidia has made it quite clear to us, that this issue only occurs when you stray from the specified memory lists available from MSI and nVidia. >>
Then when you get down to the tests he states:
<< For the memory, we used standard, run of the mill Crucial/Micron PC-2100 DDR SDRAM (Non-ECC, Non-Registered). >>
Well that memory module is on nVidia's compatibility list so why is this testbed experiencing the problem?
Then he says:
<< Turns out, the problem is that when you use dual-sided memory the second slot (slot B) of the second DDR memory bank of any given nForce motherboard, and you will consistently see a 20-25% drop in memory performance. >>
But nowhere does he mention how he came to this conclusion or why that would be a problem - just flatly blames double-sided ram.
nVidia's technical briefs on the nForce's technology and TwinBank implementation doesn't mention anything about Superstability which I find suspect.:disgust:
NSF4 mentioned Overclockers.com's review of MSI's K7N420 Pro as another experiencing this memory issue. He's using 2 identical 256MB sticks of NANYA-chipped Apacer PC2100 CAS2 for testing. So he has "strayed" from the list, so according to GamePC, that explains (albeit in another blackbox kind of way) why he's getting the performance hits.
I've got 3 sticks of Crucial PC2100 DDR SDRAM that I got for $30 each a couple of months ago. They are listed on nVidia's compatibility list. That hurdles one stumbling block: 1) not activating SS and lowering the FSB rate. The only setback I can see with "compatible" memory modules is the other issue I haven't talked about yet: 2) the twinBank arrangement limits how much memory is transferred at 128 bit and how much by 64 bit. The bank with the least amount of total memory dictates how much from each channel will be 128bit and the rest at 64bit.
So I'll have 2x256 delivering at 128bit and 256 at 64bit and all at 266MHz.