View Full Version : BIOS Optimization
BIOS Tuning: Maximum Power (http://www4.tomshardware.com/howto/20010725/index.html)
I have a question...
I have a setting in my BIOS called "Memory Remapping" and is Enabled by default. It says that it loads PCI devices above 4GB of memory and says it should not be disabled unless user is certain that the installed OS supports over 4GB or memory.
Here is my main question:
1. I don't have 4GB of memory installed for one thing
2. Windows XP supports up to 4GB of memory, not over.
Should this field be enabled or disabled in my BIOS?
And one other thing:
I think I know now why I never saw a difference setting the AGP aperture to various settings...
Maybe this is why? (http://www4.tomshardware.com/howto/20010725/bios_tuning-17.html)
This is what I'm talking about...
PresterJohn
09-07-2003, 9:02 AM
Originally posted by NDC
1. I don't have 4GB of memory installed for one thing
2. Windows XP supports up to 4GB of memory, not over.
Should this field be enabled or disabled in my BIOS?
it should be disabled...i would say the bios probably disables it automatically anyway since you have <4gb.
n7vxj
09-07-2003, 10:09 AM
:confused: NDC, and some of you wiser ones, please explain this to me!! After going to the link from NDC, and reading the article,I went into my bios, and did some changing of things.I took my latency from 2.5 to 2. I disabled the onboard sound, that I wasn't using, and disabled com 1, and com 2, since I couldn't seem to find them being used by anything. Now, all of a sudden, dnet appears to be running faster.By doing this stuff, did I ease up some of the burden on the CPU, so that it has more power for rc-5? One other thing on the latency.When I went from 2.5 to 2, did I decrease the memory slacking off time? Please explain this one to me!!!
Wizzard~Of~Ozz
09-07-2003, 11:20 AM
This is from what I've read, It is probably contradictory to what everyone is led to believe.
If you can, enable 4 way bank interleve and run at a higher CAS. lowering the CAS often results in instable behaviour, and only offers about 2% improvement, 4 way interleve offers about 10% improvement and has no negative impact on stability, so if you enable 4 way, bump the CAS up to 2.5 and run at 2% higher clock speed you will have a more stable system that's much faster.
Just something I've read in a couple of places.
CAS is the amount of time it takes to write/read memory. What it stands for (this is from top of head)
CAS Column Access Strobe, If you are familiar with matrix's (think of a sheet of grid paper) the CAS is how long it takes to access the specific column of the grid paper, this is also associated with RAS (row Access strobe) however the memory seems to be longer then it is wide making CAS have more of an impact on performance and RAS having little to none.
Don't know if this helps, if it doesn't then let me know and I'll look for a link to more information on the subject.
edit: from what I remember, 4way interleve quaters the memory, so it can start the strobe within 25% of the location as opposed to randomly. This means it will have an effect equivilent in comparison to CAS 2 of CAS .5 making it that much faster without having to run out of spec on memory and without having to worry about random memory errors.
n7vxj
09-07-2003, 11:48 AM
Wiz, I already have the 4-way interleave enabled.It just has me baffled as to why going from 2.5 to 2.0 would increase speed.From what you wrote, I assume this means I have shortened my write/read speed time.Also, I guess my disabling all that stuff, gave the cpu a little less stuff to deal with, hence freeing up more cpu power for rc-5. Please correct me if I am wrong.Did the same to my Son's box, and it seems to run rc-5 a bit faster as well.On it 4-way was disabled, so I enabled it.I'm also thinking if I disable all that power management crap, it will free up some more cpu cycles for rc-5.I guess I will just have to try it, and see!
I believe it has to do with the memory setting that gave you the performance increase in RC5. As for disabling the devices you mentioned above did not have any difference in RC5 performance increase. You will see that your CPU performance hasn't increased if you try running a benchmark on the CPU
timwei
11-03-2003, 4:02 PM
here's my take on the 2.5 to 2 speed up. The reason why you have such a significant speed up is simply in the bus. Since memory accesses usually take up 100 to 200 cpu cycles, when you decrease the memory access time from 2.5 to 2, you're increasing the bus speed between the memory and the cpu. the frequency, of course is 1/2.0 vs 1/2.5...obviously, the 1/2 is much faster.
If your memory can support such high bus speeds then you definitely shoudl have it enabled because you've probably been underclocking your memory, however if it cannot support the high speeds, then you can end up with lots and lots of memory faults.
timwei
11-03-2003, 4:07 PM
as for the memory mapping feature above 4 GB,
here's what's happening:
Memory mapping is used for pci devices as a means for I/O. Every pci device in your system is mapped to certain addresses in your RAM. So, when programs want to talk to those I/O devices, they simply do a memory read/write to those addresses assigned to your PCI devices.
The optimization here, is simply a matter of probability. Most programs will use the lower portions of your memory addresses and thus if you map all the i/o addresses above 4 GB, then you will have more readily accessible RAM for your running processes, however you will only see a difference if actually run processes that take up memory in excess of 4 GB which, according to today's home user, is very unlikely unless you're doing video editting or the like.
vBulletin® v3.7.3, Copyright ©2000-2008, Jelsoft Enterprises Ltd.