Path: csiph.com!eternal-september.org!feeder.eternal-september.org!aioe.org!.POSTED!not-for-mail From: "Rod Pemberton" Newsgroups: alt.os.development Subject: Re: Memory test? Date: Sun, 27 Dec 2015 01:00:47 -0500 Organization: Aioe.org NNTP Server Lines: 118 Message-ID: References: NNTP-Posting-Host: n4wpt9zq8xR26Ttf9mo2BA.user.speranza.aioe.org Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit X-Complaints-To: abuse@aioe.org User-Agent: Opera Mail/12.16 (Linux) X-Notice: Filtered by postfilter v. 0.8.2 Xref: csiph.com alt.os.development:9043 On Sat, 26 Dec 2015 16:15:48 -0500, Bill Cunningham wrote: > Is there a memory test you can use to check your memory sticks? I > have two DIMMs. An AMD 3500+ 64 bit processor. I have linux and can > use nasm or debug with a DOS6.21 on a virtual machine. Anyone know if > your computer going on and off could be simply bad RAM sticks or bad > code somewhere. I don't know RAM addresses. Some PCs have an advanced BIOS with a built-in machine tester, e.g., Dell PC's have a full suite of hardware tests including probably dozen different memory tests. Otherwise, you'll probably have to download a bootable CD-ROM image and burn it. E.g., Passmark's Memtest86 for Windows. You could check out Pyropus memtester for Linux. Apparently, you can activate a memory test in the kernel for some Linuxes: http://raid6.com.au/~onlyjob/posts/MEMTEST_explained/ Depending on the Linux version, you may or may not have full access to memory. See mmap() manpage. You could code a small C program for DOS. 16-bit app would be limited to low 1MB. 32-bit DPMI app could test up to 4GB. You'd need 64-bit mode code to test more memory than that with DOS. You might consider dropping a post to comp.os.msdos.* groups, like comp.os.msdos.misc or comp.os.msdos.programmer to see if anyone recalls such an app via the old DOS archives. If they do recall, and they can't find it, I might be able to. First, if you're over-clocking, turn that off, i.e., reset the BIOS settings. You might be over-clocking without realizing it, e.g., if you set any bus or memory or microprocessor speeds manually. See if the system is stable without it. If so, the over-clocking is the problem, i.e., probably insufficient cooling. Laptops can have overheat problems too because they're so enclosed. Also, many recent motherboard designs put the memory chips adjacent to the microprocessor for faster bus speeds. Unfortunately, they sometimes direct half the exhaust path for the processor's heatsink directly onto the memory cards, instead of perpendicular to the memory cards. So, you have hot air and dusty air blowing onto your memory chips ... This causes the fine dust, like that which sticks on PC's fan blades, to accumulate on the memory cards, which can cause memory errors. I'm having to clean the memory chips in this PC every six months ... Normality is never having to clean the memory chips. So, if you're mostly convinced it's a memory error, the very first thing I'd do is ground myself to not have any static charge, then power down the PC, turn off the master power switch too, then pop out your memory cards, and thoroughly blow all the dust off of them. You might want to do this over the bathtub. DO NOT do anything to cause static electricity to build up, i.e., no brushes or cloths. It's acceptable, but ill-advised for home use, to run them under water. If you do this, you *MUST* make sure they're thoroughly dry *BEFORE* applying power. I.e., use a compressed air tank or pressurized air can to blow them dry. You can generally blow the dust off with your mouth, but you *CANNOT* do so with the water. It "hides" under chips, etc, and can't be removed without high pressure air or heat. Heating electronics with a hair dryer/heat gun is acceptable too. Electronics usually go through a reflow oven or infrared surface mount oven to solder them around 400 degrees Fahrenheit. Heat to solder, wash to remove flux, blow dry with compressed air was once standard industrial practice. The next thing I'd do, if the machine is over 5 years old, is to remove the microprocessor heatsink. Get some rubbing alcohol, naptha, mineral spirits etc and remove the heatsink grease from the processor and the heatsink. This is messy. It's not as bad as getting vasoline on your hands, but still somewhat difficult to remove. You can try a product like "Goo Gone" or something with xylene. Also, have some papertowels handy. Avoid touching the grease, if possible. Put some new heatsink grease on the heatsink. Make sure it covers the entire contact area between it and the processor. Replace the processor making sure it is oriented correctly and seated flat/level. Usually, there are a few pins that must be oriented to a specific location. Do not attempt to force the processor. Make sure it's in correctly. Most modern machines have a lever activated pin grid clamping mechanism. Reattach the heat sink. Older motherboards were soldered, push in, or simply held by the processor heatsink. You can buy the heatsink grease at Radio Shack (US), electronics supply stores, better quality computer stores, or mail-order/online electronics distributors. Finally, check for any flaky or non-functional hardware. E.g., if a fan isn't spinning, it's 1) dead 2) needs bearing lubrication (silicone) or 3) something may be wrong with your power supply, i.e. intermittent, failing, damaged and unregulated, etc. If it's a power supply issue, you may see similar intermittent operation with other devices, e.g., CD-ROM. I.e., as the device warms up, power-ups, and draws more power, the failing power supply goes down. If a device is electrically damaged, you can sometimes have the same issue. If you determine it is a damaged power supply, the best choice, despite the loss of value, is to discard everything the machine, after you've backed up as much data as possible. If the hardware was damaged, it can damage a new machine. If it's a failing video card, e.g., too hot and BGA chip unsoldered, you'll usually see scrambled video or color "snowflakes" etc. Sometimes these can be salvaged by reheating the chips, e.g., heat gun, hair dryer, home oven, industrial reflow, etc. This likely means your case interior isn't being cooled properly. Modern vertical PC cases generally have two or three fans. A fan in the power supply. A fan on the processor heatsink. Sometimes a fan in front. You should also have a fan on the processor heatsink and a fan on any high-performance video card. Of course, this doesn't apply to laptops. If you're over-clocking, you'll probably need a substantial heat sink (water) on the processor, and cooling on chips that weren't heat sinked by the manufacturer, e.g., video, north/south-bridge, etc. Rod Pemberton -- The idea that sentient beings can be suppressed by a few simple rules is a farce. Even so, Isaac Asimov posited such rules for sentient artificial intelligence.