Would old Bios cause memory problems?
in [Computer]
|
Prev: Question about Excel
Next: Need help with Ipod.
From: Paul on 19 Jul 2010 23:32 Colin Trunt wrote: > "Paul" <nospam(a)needed.com> wrote in message > news:i22pn9$lc1$1(a)news.eternal-september.org... >> Colin Trunt wrote: >>> "Colin Trunt" <colin(a)trunt.com> wrote in message >>> news:Hq31o.2$Dl2.1(a)newsfe17.ams2... >>>> This relates to my earlier threads about memory problems, >>>> the seller of the ram thinks I should upgrade my Bios and it might >>>> fix the problem. >>>> >>>> >>>> I think this is highly unlikely and it's just an excuse for bad memory. >>> For example I could understand if it was not recognising the memory >>> or only seeing half of it , but it is a DDR400 PC3200 1 gig module >>> the same spec as my old one, it low density non-buffered, non-ecc. >>> >>> Hard to see how a bios change could differentiate between the two. >> I don't see a strong reason for the BIOS to make a difference for this. >> >> The BIOS seems to affect some aspects of DDR2 usage, like there >> being differences in the stability of 1GB and 2GB sticks. >> >> But in this case, you have two 1GB DDR sticks, apparently both low >> density, >> and one is behaving differently than the other. I would not expect >> a BIOS update to perform a miracle here. >> >> You'd check with CPUZ, to see if the rated clock and timing values are >> being used, and that is about it. >> >> Some DDR needs more voltage to meet timing, but I think your sticks are >> 3-3-3-8 type, and only the cheesiest of those need more voltage than >> nominal. 3-3-3-8 is more or less, industry standard timing, and not an >> enthusiast grade memory. Nominal voltage is 2.5V, and you can try a bit >> more if you want. Some motherboards will already be using a bit >> of voltage boost, like 2.65V on the 2.5V slots. But that may not be >> displayed >> anywhere, as to what the designers are doing. >> >> Paul > > > Well this is the ram with the problems spec. > ----------------- > CL3 > 1024MB > Power - 2.5V > PC-3200 DDR-400 > Non-ECC Unbuffered > RoHS and CE Certified > 2048 bits Serial EEPROM > > Desktop 184-pin connection > 16pcs 64Mb x 8 bits DDR SDRAMS > > 64M x 64bits Double Data Rate SDRAMS > Dual (Double Sided) In-Line Module (DIMM) > > Meets the industry standard JEDEC Specifications > > 100% comp. with Single/Dual Channel - High/Low Density M/boards > ----------------------- > > It's difficult to see how it is 100% compatible is if won't work with my > bios!! > > > When I have checked before everything seemed to be the same. > The other working stick is also 2.5v. > > Seems to me it is simply bad memory as it does work for a while, > often several hours. > > > My bios is:- > Award BIOS Type Phoenix - Award BIOS v6.00PG > System BIOS Date 12/26/05 > > I can't find anything better, well not without paying a lot of > money to have bios update done on line which seems to be the > only way you can do it. Bios updates are normally free? > An online update could go wrong and screw my computer. > > But bottom line is it does not work, my older module worked fine > so I don't see why the new one should not and it does state 100% > compatibility, > so that;s what I expect. > > It was pretty much an end of line machine when I got is so I would not > expect anyupdates and baios should be backward compatible! > But seems just like bad memory to me anyway so it is going back, > I will just have to see what the seller says. > BIOS updates are provided by the motherboard manufacturer. On Dell/HP/Acer or the like, those companies are responsible for the BIOS themselves, because the motherboard is "OEM". An OEM motherboard is basically contract manufactured, and the company making it (Asus/ECS/Biostar/MSI) doesn't provide support for the motherboard directly. Only very occasionally, and on some older motherboards, do you find an OEM motherboard that is also being sold as a retail motherboard. In some of those cases, you can in fact use the motherboard manufacturer BIOS. But that isn't likely to be the case here. There is not normally a reason to be buying a BIOS from ESupport or similar companies. First, you look around to see where your support is coming from. If you were to buy a BIOS from ESupport, chances are there would be no way to know what was fixed in it. If this board is from an HP computer, then HP is where you'd get the BIOS. You would start with the make and model number information for the computer, go through the driver downloads, and see if a BIOS update is listed. Compare to the BIOS release you have currently. http://h10025.www1.hp.com/ewfrf/wc/document?docname=c00378480&tmp_task=prodinfoCategory&lc=en&dlc=en&cc=uk&lang=en&product=1839474 A single motherboard design, can be used in a lot of different models of computers. It helps, if you know what computer that motherboard came out of. In this example, you can see how a similar motherboard was used in a lot of different models. http://h20141.www2.hp.com/hpparts/part_where_used.aspx?mscssid=2F9CD26405074F3A812A8664064BFA15&SearchCriteria=ER901-69001 In some cases, there won't be any BIOS updates, because there is nothing to fix. The manufacturer is content with the job done in the originally released BIOS. The BIOS is not affecting how many computers are being returned by customers, or the issues are too minor to care about. So even if you trace down the proper computer download page on the HP site, that doesn't mean there is a BIOS update waiting. It's not like they're re-issuing new BIOS for the entire warranty period. Paul
From: Paul on 19 Jul 2010 23:54 Colin Trunt wrote: > "Colin Trunt" <colin(a)trunt.com> wrote in message > news:4g81o.191605$NM4.25345(a)hurricane... >> "Paul" <nospam(a)needed.com> wrote in message >> news:i22pn9$lc1$1(a)news.eternal-september.org... >>> Colin Trunt wrote: >>>> "Colin Trunt" <colin(a)trunt.com> wrote in message >>>> news:Hq31o.2$Dl2.1(a)newsfe17.ams2... >>>>> This relates to my earlier threads about memory problems, >>>>> the seller of the ram thinks I should upgrade my Bios and it might >>>>> fix the problem. >>>>> >>>>> >>>>> I think this is highly unlikely and it's just an excuse for bad memory. >>>> For example I could understand if it was not recognising the memory >>>> or only seeing half of it , but it is a DDR400 PC3200 1 gig module >>>> the same spec as my old one, it low density non-buffered, non-ecc. >>>> >>>> Hard to see how a bios change could differentiate between the two. >>> I don't see a strong reason for the BIOS to make a difference for this. >>> >>> The BIOS seems to affect some aspects of DDR2 usage, like there >>> being differences in the stability of 1GB and 2GB sticks. >>> >>> But in this case, you have two 1GB DDR sticks, apparently both low >>> density, >>> and one is behaving differently than the other. I would not expect >>> a BIOS update to perform a miracle here. >>> >>> You'd check with CPUZ, to see if the rated clock and timing values are >>> being used, and that is about it. >>> >>> Some DDR needs more voltage to meet timing, but I think your sticks are >>> 3-3-3-8 type, and only the cheesiest of those need more voltage than >>> nominal. 3-3-3-8 is more or less, industry standard timing, and not an >>> enthusiast grade memory. Nominal voltage is 2.5V, and you can try a bit >>> more if you want. Some motherboards will already be using a bit >>> of voltage boost, like 2.65V on the 2.5V slots. But that may not be >>> displayed >>> anywhere, as to what the designers are doing. >>> >>> Paul >> >> Well this is the ram with the problems spec. >> ----------------- >> CL3 >> 1024MB >> Power - 2.5V >> PC-3200 DDR-400 >> Non-ECC Unbuffered >> RoHS and CE Certified >> 2048 bits Serial EEPROM >> >> Desktop 184-pin connection >> 16pcs 64Mb x 8 bits DDR SDRAMS >> >> 64M x 64bits Double Data Rate SDRAMS >> Dual (Double Sided) In-Line Module (DIMM) >> >> Meets the industry standard JEDEC Specifications >> >> 100% comp. with Single/Dual Channel - High/Low Density M/boards >> ----------------------- >> >> It's difficult to see how it is 100% compatible is if won't work with my >> bios!! >> >> >> When I have checked before everything seemed to be the same. >> The other working stick is also 2.5v. >> >> Seems to me it is simply bad memory as it does work for a while, >> often several hours. >> >> >> My bios is:- >> Award BIOS Type Phoenix - Award BIOS v6.00PG >> System BIOS Date 12/26/05 >> >> I can't find anything better, well not without paying a lot of >> money to have bios update done on line which seems to be the >> only way you can do it. Bios updates are normally free? >> An online update could go wrong and screw my computer. >> >> But bottom line is it does not work, my older module worked fine >> so I don't see why the new one should not and it does state 100% >> compatibility, >> so that;s what I expect. >> >> It was pretty much an end of line machine when I got is so I would not >> expect anyupdates and baios should be backward compatible! >> But seems just like bad memory to me anyway so it is going back, >> I will just have to see what the seller says. > > > Anyway I jsut stuck the bad ram back in, this time in slot 3 as unmatched. > however as it wil not work by itself it is a vain hope it will work like > this!! > > This is some CPUZ stuff for the memory info etc... > > Chipset > ------------------------------------------------------------------------- > > Northbridge ATI Xpress 200 (RS480) rev. 10 > Southbridge ATI SB400 rev. 00 > Graphic Interface PCI-Express > PCI-E Link Width x16 > PCI-E Max Link Width x16 > Memory Type DDR > Memory Size 2048 MBytes > Channels Single > Memory Frequency 199.0 MHz (CPU/10) > CAS# latency (CL) 3.0 > RAS# to CAS# delay (tRCD) 3 > RAS# Precharge (tRP) 3 > Cycle Time (tRAS) 8 > Bank Cycle Time (tRC) 11 > DRAM Idle Timer 16 > Command Rate (CR) 2T > > Memory SPD > ------------------------------------------------------------------------- > > DIMM # 1 > SMBus address 0x50 > Memory type DDR > Manufacturer (ID) (0000000000000000) > Size 1024 MBytes > Max bandwidth PC3200 (200 MHz) > Part number 128M64K-40C > Serial number 040413C3 > Number of banks 2 > Data width 64 bits > Correction None > Registered no > Buffered no > Nominal Voltage 2.50 Volts > EPP no > XMP no > JEDEC timings table CL-tRCD-tRP-tRAS-tRC @ frequency > JEDEC #1 2.0-2-2-6-n.a. @ 133 MHz > JEDEC #2 2.5-3-3-7-n.a. @ 166 MHz > JEDEC #3 3.0-3-3-8-n.a. @ 200 MHz > > DIMM # 2 > SMBus address 0x52 > Memory type DDR > Manufacturer (ID) (FFFFFFFFFFFFFFFF) > Size 1024 MBytes > Max bandwidth PC3200 (200 MHz) > Part number > Serial number FFFFFFFF > Manufacturing date Week 255/Year 255 > Number of banks 2 > Data width 64 bits > Correction None > Registered no > Buffered no > Nominal Voltage 2.50 Volts > EPP no > XMP no > JEDEC timings table CL-tRCD-tRP-tRAS-tRC @ frequency > JEDEC #1 2.0-2-2-6-n.a. @ 133 MHz > JEDEC #2 2.5-3-3-7-n.a. @ 166 MHz > JEDEC #3 3.0-3-3-8-n.a. @ 200 MHz > > DIMM # 1 > SPD registers > 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F > 00 80 08 07 0D 0B 02 40 00 04 50 70 00 82 08 00 01 > 10 0E 04 1C 01 02 20 C0 60 70 75 75 3C 28 3C 28 80 > 20 60 60 40 40 00 00 00 00 00 37 46 28 28 50 00 00 > 30 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 A8 > 40 00 00 00 00 00 00 00 00 00 31 32 38 4D 36 34 4B > 50 2D 34 30 43 00 48 59 00 00 00 00 00 00 00 00 04 > 60 04 13 C3 00 00 00 00 00 00 00 00 00 00 00 00 00 > 70 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > 80 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > 90 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > A0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > B0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > C0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > D0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > E0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > F0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > > DIMM # 2 > SPD registers > 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F > 00 80 08 07 0D 0B 02 40 00 04 50 70 00 82 08 00 01 > 10 0E 04 1C 01 02 20 C0 60 70 75 75 3C 28 3C 28 80 > 20 60 60 40 40 00 00 00 00 00 37 46 28 28 50 00 00 > 30 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 A8 > 40 FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF > 50 FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF > 60 FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF > 70 FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF > 80 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > 90 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > A0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > B0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > C0 00 00 00 00 00 00 00 00 E1 00 00 00 00 00 00 00 > D0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > E0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > F0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > > > So that all seems identical to me. > > I am tempted to leave it in and give it 'one last chance' - lol. > However when it reboots it tends to leave the drive 'dirty', > I might give it one last run on memtest in this configuration. > > > One thing which does catch my eye is all the FF's in the > SPD registers on DIMM#2 (the 'bad memeory') however > I have no idea what that implies! The implication is printed in CPUZ. Part number <--- no part number Serial number FFFFFFFF <--- bogus Serial Number Manufacturing date Week 255/Year 255 <--- bogus manuf. date DIMM #1 has a serial number, meaning it is possible all sticks have their own number. Some Crucial memory I bought is like that. DIMM #2 still has some info stored in it, but a part of it was filled with FF's. There is a length field defined, and if they wanted to shorten the declared "valid" area, they could have changed the length. The SPD EEPROM can come with all manner of mistakes in it. The worst I've seen, is where the SPD declares the wrong memory size for the module. The BIOS was still able to figure out the correct size, due to an older probing technique still included in the BIOS. So that is not the worst table I've ever seen. Another error that cannot be forgiven, is when the SPD contains bad timing information. That can prevent the computer from posting, and would cause a lot of product returns. A double check, is to compare the information in the SPD table, versus the chip part number, to see if the declaration is consistent with the part number. If you look up the memory chip, and a certain dash extension on the part number tells you it is DDR400 CAS3, then you'd expect to see the timing table list DDR400 CAS3 as the highest speed entry in the table. Enthusiast memory is made, by bumping up the voltage, lowering settings like CAS to CAS2, and re-testing on the production line. Such a part, will come with an assurance on the printing on the package that it is DDR400 CAS2, but the SPD may contain DDR400 CAS3. Such memory is not fully meeting its specification, until the customer dials in the settings manually. Such a trick is necessary, in order that the first time you install the stick, the "relaxed" DDR400 CAS3 value is used automatically by the BIOS, guaranteeing the computer can computer POST and boot. There was a time, when the more aggressive value was stored in there, and then the computer wouldn't start. Your stick of RAM is not an enthusiast product, so you have nothing to worry about in that regard. It looks like it is intended to work properly without any adjustments. Only if you wanted to keep the stick, would you contemplate turning up the VDimm voltage. If a couple of passes of memtest only yielded one error, that might be correctable by a voltage adjustment. In your case, having the computer reboot, doesn't sound nearly as encouraging (i.e. voltage adjustment might not be enough, or the stick might start failing more frequently as time passes). The motherboard itself could be bad, but then, you can always do some testing with Prime95 (mersenne.org/freesoft). There is a stress test option in that program, which helps identify weak computer components. One hour of Prime95, is equal to the motherboard sitting idle in Windows for 100 hours. So Prime95 can be used as an accelerated test of computing health. If your VDimm or VCore on the motherboard was weak, that might help uncover it. If that were the case, the program might report failures with either stick used independently. If the failures still follow the "bad" stick, then the bad stick is the guilty party. Paul
From: Colin Trunt on 20 Jul 2010 00:17 "Paul" <nospam(a)needed.com> wrote in message news:i235cf$dde$1(a)news.eternal-september.org... > Colin Trunt wrote: >> "Paul" <nospam(a)needed.com> wrote in message >> news:i22pn9$lc1$1(a)news.eternal-september.org... >>> Colin Trunt wrote: >>>> "Colin Trunt" <colin(a)trunt.com> wrote in message >>>> news:Hq31o.2$Dl2.1(a)newsfe17.ams2... >>>>> This relates to my earlier threads about memory problems, >>>>> the seller of the ram thinks I should upgrade my Bios and it might >>>>> fix the problem. >>>>> >>>>> >>>>> I think this is highly unlikely and it's just an excuse for bad >>>>> memory. >>>> For example I could understand if it was not recognising the memory >>>> or only seeing half of it , but it is a DDR400 PC3200 1 gig module >>>> the same spec as my old one, it low density non-buffered, non-ecc. >>>> >>>> Hard to see how a bios change could differentiate between the two. >>> I don't see a strong reason for the BIOS to make a difference for this. >>> >>> The BIOS seems to affect some aspects of DDR2 usage, like there >>> being differences in the stability of 1GB and 2GB sticks. >>> >>> But in this case, you have two 1GB DDR sticks, apparently both low >>> density, >>> and one is behaving differently than the other. I would not expect >>> a BIOS update to perform a miracle here. >>> >>> You'd check with CPUZ, to see if the rated clock and timing values are >>> being used, and that is about it. >>> >>> Some DDR needs more voltage to meet timing, but I think your sticks are >>> 3-3-3-8 type, and only the cheesiest of those need more voltage than >>> nominal. 3-3-3-8 is more or less, industry standard timing, and not an >>> enthusiast grade memory. Nominal voltage is 2.5V, and you can try a bit >>> more if you want. Some motherboards will already be using a bit >>> of voltage boost, like 2.65V on the 2.5V slots. But that may not be >>> displayed >>> anywhere, as to what the designers are doing. >>> >>> Paul >> >> >> Well this is the ram with the problems spec. >> ----------------- >> CL3 >> 1024MB >> Power - 2.5V >> PC-3200 DDR-400 >> Non-ECC Unbuffered >> RoHS and CE Certified >> 2048 bits Serial EEPROM >> >> Desktop 184-pin connection >> 16pcs 64Mb x 8 bits DDR SDRAMS >> >> 64M x 64bits Double Data Rate SDRAMS >> Dual (Double Sided) In-Line Module (DIMM) >> >> Meets the industry standard JEDEC Specifications >> >> 100% comp. with Single/Dual Channel - High/Low Density M/boards >> ----------------------- >> >> It's difficult to see how it is 100% compatible is if won't work with my >> bios!! >> >> >> When I have checked before everything seemed to be the same. >> The other working stick is also 2.5v. >> >> Seems to me it is simply bad memory as it does work for a while, >> often several hours. >> >> >> My bios is:- >> Award BIOS Type Phoenix - Award BIOS v6.00PG >> System BIOS Date 12/26/05 >> >> I can't find anything better, well not without paying a lot of >> money to have bios update done on line which seems to be the >> only way you can do it. Bios updates are normally free? >> An online update could go wrong and screw my computer. >> >> But bottom line is it does not work, my older module worked fine >> so I don't see why the new one should not and it does state 100% >> compatibility, >> so that;s what I expect. >> >> It was pretty much an end of line machine when I got is so I would not >> expect anyupdates and baios should be backward compatible! >> But seems just like bad memory to me anyway so it is going back, >> I will just have to see what the seller says. >> > > BIOS updates are provided by the motherboard manufacturer. On > Dell/HP/Acer or the like, those companies are responsible for > the BIOS themselves, because the motherboard is "OEM". An OEM > motherboard is basically contract manufactured, and the company > making it (Asus/ECS/Biostar/MSI) doesn't provide support for > the motherboard directly. > > Only very occasionally, and on some older motherboards, do you > find an OEM motherboard that is also being sold as a retail > motherboard. In some of those cases, you can in fact use the > motherboard manufacturer BIOS. But that isn't likely to be > the case here. > > There is not normally a reason to be buying a BIOS from ESupport > or similar companies. First, you look around to see where your > support is coming from. If you were to buy a BIOS from ESupport, > chances are there would be no way to know what was fixed in it. > > If this board is from an HP computer, then HP is where you'd get the BIOS. > You would start with the make and model number information for the > computer, go through the driver downloads, and see if a BIOS update is > listed. Compare to the BIOS release you have currently. > > http://h10025.www1.hp.com/ewfrf/wc/document?docname=c00378480&tmp_task=prodinfoCategory&lc=en&dlc=en&cc=uk&lang=en&product=1839474 > > A single motherboard design, can be used in a lot of different > models of computers. It helps, if you know what computer that > motherboard came out of. In this example, you can see how a similar > motherboard was used in a lot of different models. > > http://h20141.www2.hp.com/hpparts/part_where_used.aspx?mscssid=2F9CD26405074F3A812A8664064BFA15&SearchCriteria=ER901-69001 > > In some cases, there won't be any BIOS updates, because there is > nothing to fix. The manufacturer is content with the job done > in the originally released BIOS. The BIOS is not affecting > how many computers are being returned by customers, or the > issues are too minor to care about. So even if you trace > down the proper computer download page on the HP site, that > doesn't mean there is a BIOS update waiting. It's not like > they're re-issuing new BIOS for the entire warranty period. > > Paul Everest leads me here for bios updates http://biosagentplus.com/?ref=40 For product (bios) info it goes here http://www.phoenix.com/en/products/default.htm (page not found - very helpful) email address for product support here, but it looks like they are only interested in sellinlg new stuff, however I can try it anyway. The biosagent link give a number of updates for various things however when you tryy and download an update it lead to 'join now' page for $30 a year and tell you nothing else. SO I do not want to pay $30 for something which probably will make no different and could possible leave my computer unusable if something goes wrong during the update, so I think the best option is to try return the memory for a refund. http://www.phoenix.com/pages/support--3 I seem to go roundin circles looking for bios updates at HP. It says to click the bios link on another page. http://h10025.www1.hp.com/ewfrf/wc/softwareCategory?os=228&lc=en&cc=uk&dlc=en&sw_lang=&product=1839474#N1949 And there is not one. I have googled the bios name and verions and 'update' but it get me nowhere just other people threads on forums with the same problem!!
From: Paul on 21 Jul 2010 01:03 Colin Trunt wrote: > > > Well I have some significant results to report on the memory channels > confusion. > I put one gig in slot one and one gig in slot two and cpuz reported dual > channel!! > I also tried 1 in 1 and 1 in slot 3 but that came back single channel so at > least I know > for sure what my channels for matched memory is. > > I also ran the memeory benchmark in Everest, which produced some inpressive > results, almost doubling the memory through put. > > In fact it came up faster than the 4800 X2 on both the read and the writes, > also > it came up faster than the FX 51 on at least one of them. > > On single channel it is only around the AMD 3200+ mark, thats about 2900 > MB/s > for the reads, on the dual channel it was getting about 5400 or more. > > Those are impressive figures but I am not sure how they relate to normal > operation as > opposed to a benchmark test. > > Anyway I am back to my old config wiith 256 meg in slot 1 and 1gig in slot > 2, that's perhaps > not a recommended configuration but I have been running like that for years > without any apparent > problems so I will stick with it. > > So your setup could look like this. DIMMs in 1 and 3 would be on the same channel, and matched DIMMs in 1 and 2 would be dual channel. Channel Channel DIMM1 DIMM2 DIMM3 DIMM4 Since you've put the 256MB in DIMM1 and the 1GB in DIMM2, that is one electrical load per channel. But likely it's running in virtual single channel mode, with reduced memory bandwidth (in synthetic benchmarks). Channel Channel 256MB 1GB --- --- The real world difference between the two cases, isn't quite as marked, as the benchmark results would suggest. It isn't directly proportional (i.e. double bandwidth doesn't mean double the execution speed - it is only slightly faster). Running like this, is single channel as well. But since there are two electrical loads on one channel, the BIOS may drop the clock rate to the RAM, when running this way. This could end up being slower than the previous setup. Channel Channel 256MB --- 1GB --- You can use something like SuperPI, to test how much of a difference, some change in memory setup is making. The "1 million digit" test is good for a quick check, on processors with small caches. If the processor has a large cache (like some of the new six core processors), then the "32 million digit" test might be a fairer test to run. Using a larger memory footprint on a test, helps ensure the processor cache doesn't provide an unfair advantage. http://www.techpowerup.com/downloads/366/Super_PI_Mod_v1.5.html Paul
From: Colin Trunt on 21 Jul 2010 14:44
"Paul" <nospam(a)needed.com> wrote in message news:i25v2k$npq$1(a)news.eternal-september.org... > Colin Trunt wrote: > >> >> >> Well I have some significant results to report on the memory channels >> confusion. >> I put one gig in slot one and one gig in slot two and cpuz reported dual >> channel!! >> I also tried 1 in 1 and 1 in slot 3 but that came back single channel so >> at least I know >> for sure what my channels for matched memory is. >> >> I also ran the memeory benchmark in Everest, which produced some >> inpressive >> results, almost doubling the memory through put. >> >> In fact it came up faster than the 4800 X2 on both the read and the >> writes, also >> it came up faster than the FX 51 on at least one of them. >> >> On single channel it is only around the AMD 3200+ mark, thats about 2900 >> MB/s >> for the reads, on the dual channel it was getting about 5400 or more. >> >> Those are impressive figures but I am not sure how they relate to normal >> operation as >> opposed to a benchmark test. >> >> Anyway I am back to my old config wiith 256 meg in slot 1 and 1gig in >> slot 2, that's perhaps >> not a recommended configuration but I have been running like that for >> years without any apparent >> problems so I will stick with it. > > So your setup could look like this. DIMMs in 1 and 3 would be on the > same channel, and matched DIMMs in 1 and 2 would be dual channel. > > Channel Channel > > DIMM1 DIMM2 > DIMM3 DIMM4 > > Since you've put the 256MB in DIMM1 and the 1GB in DIMM2, that is > one electrical load per channel. But likely it's running in virtual > single channel mode, with reduced memory bandwidth (in synthetic > benchmarks). > > Channel Channel > > 256MB 1GB > --- --- > > The real world difference between the two cases, isn't quite as marked, > as the benchmark results would suggest. It isn't directly proportional > (i.e. double bandwidth doesn't mean double the execution speed - it is > only slightly faster). > > Running like this, is single channel as well. But since there are > two electrical loads on one channel, the BIOS may drop the clock > rate to the RAM, when running this way. This could end up being slower > than the previous setup. > > Channel Channel > > 256MB --- > 1GB --- > > You can use something like SuperPI, to test how much of a difference, > some change in memory setup is making. The "1 million digit" test > is good for a quick check, on processors with small caches. If the > processor has a large cache (like some of the new six core processors), > then the "32 million digit" test might be a fairer test to run. > Using a larger memory footprint on a test, helps ensure the > processor cache doesn't provide an unfair advantage. > > http://www.techpowerup.com/downloads/366/Super_PI_Mod_v1.5.html > > Paul Well I have donr some tests on the benchmark speeds using Everest and it gives] In slots 1 and 2 (256meg in one) 3005 mbs read 1102 mbs write 68.2 latnecy In slots 1 and 3 (256meg in one) 3015 mbs read 1106 mbs write 67.1 latnecy So a very slight improvement but noting significant really I got a slower result just now but I have more things running so it appears to make no real difference. OK I got the return address from the retailer today so I guess I will be sending it back, I feel a little guilty I have not tried updating the bios but I do think it is unlikely to fix it for various reasons. Anyway he describes the memory as 100% compatibe, nothing on his page about BIOS updates!! I am still tempted to have a go at the bios update out of curiosity. |