Discussion on:

Five lesser-known problems that can slow down a PC

maybe it's me, but all but the system board drivers appear to be pretty common, not at all lesser known. it has, to me at any rate, been obvious that disk thrashing shows a lavk of ram for what is being asked of the system. the mobo drivers i rarrely think of myself, so yeah, that's pretty uncommon. but the rest? posibly OVERLOOKED, but certainly not lesser known.

Doesn't the processor turn out to be one of the culprits in a slow PC? If the PC has 64 RAM, but has an old 133Hz processor, the processor could be the culprit. Am i right?

I agree with kiran_38, that it could be a slower or incompatible processor. At least that's one other thing that slowed mine down. But I've noticed also that if there are viruses lying resident in files and folders, this causes not only slowdowns, but also constant freezing of your processes, or the "blue screen of death". I have cable modem, so this is a continuous problem for me. The solution to this is to scan my system daily to rid it of the viruses and keep my virus definitions updated.

Another culprit, I have found is advertisers loaded in the system.

Programs such as Morpheus, Gator, Buy-Now, On-Now, etc. are constantly sending, or trying to send,information back to their owners.

I also run Ad-Aware (www.downloads.com). This is a freeware utility that will remove this from a system.

After running Ad-Aware, I have seen a tremendous increase in speed on the machine.

John

Most people forget to close ports using firewall software and not monitor Denial of service attacks they can bring computer to a halt in matter of minutes. I was running a system using IIS sever and had an instructor in college told him that my system was secure he could not hack in he set a denial of sevice attack at every port while I was trying to close the ports I forgot it sap the computer resources and let in a torgan and he had full control of my computer the lesson I learn is if you are getting attack disconect it from the network is the only reliable option and regroup better run to fight again than to be taken over and no control

Not a flame, just a piece of advice.
Please! For the sake of your reputation, not to mention your readers, use some punctuation, & check your spelling.

EDIT: should have rechecked MY spelling before posting

For the most part, the processor is the least likely problem in slow performance. Your slowest points are almost always due to inadequate RAM and slow hard drive speed. The processor is very rarely rnning at max output unless you are using graphics a great deal (i.e. gaming) and even then, the RAM and video card are more prone to cause slow response.

that is true. the name of o the game is to have what I call the "Three Part Package":
1) Lots of memory 64/256/512 or memory
2) Powerful processor P4, AMD Atheon
3) A large hardrive anything over 30 Gigs for home use

If you can afford to have that combination you will not have much of any problems at all. Of course the combination will change over time because of the technology change in our hardware and software usage.

if after cleaning the system of spywares virus and startup programs the system continues slow check this:
first of all look at the mainborad.does it have an amr slot?remove anything connected to it,these nasty cards can really slow down a pc,replace it by a "real" card if possible.look for a bios update,if you cant find it then clear the cmos and reconfigure it.check te hard drive cables to see if they have 40 or 80 wires.check the cpu temperature.install the most recent drivers from the manufacturers site.do not install softmodems.if you have a celeron you can try to overclock it but whatch the temperature.

surely a reminder of some good points is never a bad idea. stop being so haughty and remember that this forum doesnt only serve seasoned tech folk but others as well. We were all new once bud.

Good for you, ZedB1228! This forum (and others like it) are invaluable to me as I progress from "desperately seeking for answers" to "seasoned professional." I'm the one-and-only tech support without peers whose "brains I can pick" or abundant resources. I'm grateful for all the help I can get!

And thanks for ALL your articles, Brien!

Whether newbies or seasoned techs, working with MS always a good idea to not overanalyze. KISS -keep it simple stupid - especially with preconfig machines.

I am grateful for all the information provided here. I am not certified OR degreed, but I am able to comprehend quite a bit of what is discussed here and put it to good use. Amazing, huh? I think giving off a nasty attitude toward someone giving helpful advice is just not helpful in this forum. Thank you to everyone who provides all the great tips and information.

Especially virtual memory. That should be routine to check when dealing with system performance issues.

In general, what are optimum virtual memory (VM) settings? I have read it is best to set them at half of the machines physical memory, instead of letting Windows handle it. Have any of you IT pro's found it beneficial to do this, or is it better to let Windows handle it? Is this true even for Windows ME or XP?

mroddy@voyager.net made a good point further down the board. This should get you there (in theory):

Dynamic Swap File versus Fixed
http://www.techrepublic.com/forumdiscuss/thread_detail.jhtml?thread_id=100826&message_id=610366

Size really depends on the limitations of the OS and the intended use of the machine, but definitely a fixed swap file. For Windows 9x (ME), I like about 2x physical RAM; for NT (2K & XP) I like 2.5x. Keep in mind, the more VM you have, potentially, the slower the machine will be. Current RAM modules average around 6-10 nanoseconds, whereas hard disk seek times, for a good SCSI are around 5-8 milliseconds ... about 1000x slower. Having said that, there was one 2K machine I worked on that was unfortunately being used for graphic design (this wasn't my choice!): it required 6 GB of VM (scattered across multiple partitions and disks) to get the job done. It was painfully slow at times, but did provide ample opportunity for coffee/smoke breaks.

It took THAT much Virtual Memory to run graphics in 2K??? Wow, that setup is similar to my favorite toy. (Athlon 1.3, currently on 512 RAM, 60 G HD)I like to play with a 3D graphics program when I have the time. Recently I heard of the possibility of increasing the minimum page file size to match the maximum. With the maximum set to 2.5 times the amount of physical memory. The real culprit may have been DYNAMIC paging in Win 2K. Changing the file size constantly with the memory flow is what slows things down. I only set mine that way the other day and havent tested the theory, but it's worth a try.

Giving a little thought first will show you the fallacy of any such VM rule as "half of the machines physical memory"-and that's true for *any* direct ratio whether it's half, 2X or 3X or whatever.

First, the total memory required by a PC dependson what it's doing. If you can't predict that then it's foolish to limit that memory by setting a maximum size to the swap file-all that will do is generate 'out of memory' crashes when you need more. Personally I'd rather the PC slow down but keep working vs. crashing.

Second, a direct ratio assumes that a PC with less physical RAM will also need less virtual memory. Huh? Take your 'half' ratio. That says that my PC with 64MB of RAM needs only 32MB of VM while my PC with 256MB of RAM needs 128MB of VM. I'd say it's more likely to be just the opposite.

If you simply *must* apply a ratio then at least make it a reverse one. Say some figure (I like 8192) divided by the amount of physical RAM. That gives the 64MB PC 128MB of VM & the 256MB PC 32MB of VM. How did I get 8192? I determined by trial & error the amount of VM I though appropriate for the 64MB PC & worked backwards to the factor. Then I applied it to the 256MB PC to cross-check that the factor I derived seemed reasonable.

In practice I find it best to put in plenty of RAM & hard disk then let Windows handle the swap file.

If you have a machine that worked OK with say 128MB RAM and a 256MB swapfile, you might think that upgrading to 512MB with zero swapfile would be an improvement (more real RAM). but you'd be wrong. Unless you grant Windows control over the swapfilesize, my experience is that it will run slower.

You can't seem to run Windows without a swapfile (why not?).

Also - does having more memory to manage OF ITSELF increase the background memory management workload and slow it all down?

If sothere may be some optimum level of memory for a given machine's usage pattern.

It would be nice to get some authoritative answers (someone MUST know how it all works!) rather than "2 or 3 times this number" heuristics

When Windows services are idle, they are swapped out, to lessen consumption of system RAM. So, even if you had 2Gb of RAM, with 1.75Gb free, Windows would still want to assign a 32K system service to swap file after it had been idle for a certain period.

Have you ever noticed that when you wake your PC up after a period of inactivity, it thrashes the hard drives for a moment? That is Windows taking the now required services from the swapfile, and placing them back into real RAM.

My rule is, and I apply it to all my clients machines, heaps of RAM first, then a pagefile that starts at 2.5 times real RAM, and has no upper limit.

Used with Norton SpeedDisk, this allows you to have a perfectly contigous pagefile that rarely if evergrows.

Which is where I disagree with the "let Windows handle the pagefile" malarky. Great if you want a pagefile that ends up in a million pieces scattered around the hard drive. Definite performance issues will arise.

That's right; you do NOT want to let Windows manage your VM. As Abitec mentioned, you want your pagefile to be as contiguous and unfragmented as possible. Nor do you want to take the performance hit of windows having to grow its pagefile as it needsit.

And if the machine in question is one of many customer workstations, you definitely want to make sure that you specify a ?minimum? file size. Unless you have lots of time to check every customer's computer frequently, you don't want to take the chance of the pagefile's host drive becoming full. How much free drive space they have, is something most end-users aren't even aware of (or care about). So you have to ensure that an adequate amount of space is reserved for your pagefile. And as mentioned before, if you don't feel comfortable with second guessing the upper limit needs of a particular machine -- then don't specify one.

So the bottom line: Let Windows management your swapfile? Just say No!

I am a firm believer that contiguous Page/Swap files are the best practice when using MS Windows xxxxx. IMHO Norton Speed disk is the best product for defragmenting and optimizing the Page/Swap file location.

The 2.5 x Total physical RAM for thebeginning size is okay as long as your HDD has plenty of free space. Definitely much easier to set up when your PC is new. With newer motherboards and chipsets driving total addressable RAM into the 4GB range I would be inclined to set up a new system with 2.5 x Total physical RAM.

By default, Windows places the pagefile on the boot partition where the operating system is installed. To determine the size of the pagefile multiply the amount of physical RAM by 1.5 to a maximum of 4095MB. However, placing the pagefile on the bootpartition does not optimize performance because Windows has to perform disk I/O on both the system directory and the pagefile. Therefore, it is recommended that you place the pagefile on a different partition and different physical hard disk drive so that Windows can handle multiple I/O requests more quickly.

This KB can be found @
http://support.microsoft.com/default.aspx?scid=KB;en-us;q197379

Board drivers are very important with any of the versions of Windows 95, 98, ME, NT, 2000. Little known is that those drivers can actually be a henderence (Speed) if not loaded in the correct order with relation with other Hardware drivers. The board Manufactures are not so quick to give instructions on the proper squence of loading drivers. Intel trains there IPD's on the correct order of driver loading for optimum performance.
I have had this experience in the past where a machine will run wonderful until the Chipset drivers are loaded then take 1 minute or more to boot to windows. When I applied the training in a test I found that the same machine performed better with no change in boot time. What we as Tech's should do is demand this information from the Board Manufactures so we can pass along the best experience to our customers.

Board drivers are very important with any of the versions of Windows 95, 98, ME, NT, 2000. Little known is that those drivers can actually be a henderence (Speed) if not loaded in the correct order with relation with other Hardware drivers. The board Manufactures are not so quick to give instructions on the proper squence of loading drivers. Intel trains there IPD's on the correct order of driver loading for optimum performance.
I have had this experience in the past where a machine will run wonderful until the Chipset drivers are loaded then take 1 minute or more to boot to windows. When I applied the training in a test I found that the same machine performed better with no change in boot time. What we as Tech's should do is demand this information from the Board Manufactures so we can pass along the best experience to our customers.

Some folks seem to have a mis-perception related to RAM speed. The speed rating of a pieec of RAM is the maximum bus speed it is rated for, not how fast it actually runs in a given machine.

That being said, you can put higher speed RAM in a slower machine. PC133 RAM of the same form factor will work in a 100 MHz buss machine that would be just as happy with PC100 RAM. The PC133 RAM won't run at 133 by itself.

On the other hand, putting PC100 RAM in a machine with a 133 buss would likely cause difficulties because the RAM is not able to run at the higher speed. It *might* work, but you would eventually encounter memory errors.

With the advent of faster bus/memory speed, don't forget airflow around the memory area. Symptoms of bad or overheat memory ranges from slowing down of the system, to system errors

Regardless of what manufacturers say about bus speed compatibility, my experience is that PC133 DIMMS may not be recognized in a PC100 machine.

To be safe, match the bus speed.

Otherwise, I agree about RAM speed ratings. Lower RAM speeds (in nanoseconds) will run in sockets with higher RAM speed specifications.

If a PC100 machine won't recognize PC133 DIMMs then the problem is something other than the speed. Dig a little deeper into the specifications. I had an experience like that with a PC66 machine-but it turned out that it was designed for '2-clock line' DIMMs and all the new DIMMs (including newer PC66 ones) are '4-clock line' DIMMs. I don't know what the clock lines are, but that was the problem, not the speed.

By the way, you have your last statement backwards. Higher RAM speeds will run in sockets with lower RAM speed specifications.

As I refurbish old(er) machines one overlooked problem I have frequently come across is a lack of heat transfer complound between heatsink and
CPU on meany of the OEM systems.

Often a dab of compound will revive an old dog to at least a middle aged puppy!!

I was suprised to learn from a customer that the several former techs that he used did not see a need to use heat transfer compound. He just had a serious performance issue with his Win98 box running an AMD processor. He informed me that he had just gotten a new CPU installed and the tech didn't use and compound. After we added some the compound the issues resolved and the machine began to run much faster and more reliably than he had ever experienced. Just a note that some of the "old dogs" don't even use the "old tricks" to cure common problems.

It was so simple... MCSE, and the thought the Microsoft had all the answers for controllers and everything else.

WRONG. Read the books that come with your hardware. A blue screen is not a poorly written application as Microsoft Pitches most ofthe time. It commonly is a poorly executed installation of the operating system or some other facet installed improperly.

Save yourself a headache next time and install the System board disk AFTER the operating system.

Thanks for proving my point on the certifications!

processors and viri certainly, make no mistake, slow things down. but viri fall under "extraprdinary circumstances" being pretty uncommon (my helpdesk covers 20,000+ people, and it's pretty rare for us) and a slow processor should be taken into account-that is, when most people say their systems are slow, it's because they've noticed it behaving SLOWER than normal. of course a 133 is going to seem slow. to a person used to a 500+ machine. to a guy with a 66, (believe me, they are still out there) it flies. so you have to take into account the processor speed then DIScount it. The author is talking about fixing something out of the ordinary. my statement was that I really didn't think these things to look for and try were uncommon, possiblyoverlooked, but hardly a revelation

im just wondering, does the ram speed have something to do with boosting the performance of the pc? if a ram has a speed of 100mhz and has a 128mb size, is there any difference with the performance of a ram which has a speed of 133mhz and a 64mb size? how about the cpu speed? i noticed there are two cpu speeds in single cpu, they claim it has a speed of 450mhz but also claim that the same cpu has a 100mhz speed.

the second speed of 100 or 133 mhz in cpu rating is the speed of the bus that goes BETWEEN the memory and the cpu itself. the 450 is how fast things actually occur INSIDE the cpu. the actual circuits that form the pipe between the two is the 100/133. i'll tell you, i have seen a noticeable, if not significant, in going from a celereon 733/100 128meg to a p3 733/133 with 128 meg. but i'm critical and picky, so , perceived result may vary
with a difference in memory size (128vs64) i can't see that there is any advantage to the 133 memory, simply because the clock speed of one-third more than the 128 surely can't make up for the 100% more physically there.

The CPU Level-2 cache could've been a reason for the noticeable performance increase (128k vs 256k).

but man, did i notice it! i wouldn't have believed a little jump like that would have come across that way

P3s perform just like Celerons, that's why they cost more! I would be surprised if you didn't see an improvement!

Actually, in many PIII systems, the 133Mhz is the bus speed from the CPU to the Front Side Bus (FSB). The speed from the FSB to the RAM is still at 100Mhz. The 810e chipset is one that does this. It is also why you can crack open a lot of retail boxen that are using PC100 memory with a chip that runs at a 133Mhz bus speed.

So, depending on the chipset, the jump from 100 to 133Mhz FSB may no be as big a jump as expected.

I would have to comment that the virus software scan properties can really affect performance. Scanning all files, or all inbound and outbound files, plus download scans will definitely affect performance on older PC's. Many software apps are no longer run locally, therefore network traffic is much greater. We still use PI 166 pc's in our environment and have found that we need to reconfigure our virus scanning utilities to accommodate the increased traffic.

Findfast...slows everything down. I don't know why they didn't call it findslow. I'd wap it out with task manager first chance you get and then delete it from the startup. That'll make thing go a little faster. I'm not sure if indexing is any better but haven't developed a huge dislike for it yet.

Those two slow EVERYTHING down. And FindFast is a common cause of a blue screen when removing cd's you've been using...

I have found that the best way to install an MS O/S is to actually remove any PCI/ISA cards that are not essential (i.e. everything except the graphics card) and then install the O/S. Then install the MOBO drivers as the very first thing you do, then switch off and add each card one at a time. Sounds weird but since I started using this technique I have had much fewer problems with O/S installs and much stabler machines in the long run

This is not weird at all since most add-in card manufacturers tell you to do this anyway, it makes sense that it would work better. But I hadn't thought of it, so thank you.

Hmmm... that reminds me, all the certificats and what the Actually mean:

MCSE = Must Consult Someone Experienced

I have over 15 years of tech experience. Unless you have an MCSE, or any other cert. you not only *shouldn't* talk...you may be talking out your @ss...

While experience should weigh in heavier than certs when it comes to hiring...experience *plus* relevent certs should almost *always* win over experience without them.

u know...that sort of arrogance is what stopped me from getting a job for several years. I also have about 10 years tech experience, but I could never afford the certs. i mean the MCSE especially is prohibitive. My current helpdesk job they took a chance on me and it paid off. I proved my viability and experience very quickly.

So get YOUR head out of your @ss and realise there are other considerations.

TwitcH

PS I'm now getting my A+ coz I can AFFORD it.

I've never heard someone say they didn't get a cert because it costs too much. What a crock! If you have the experience that you claim, and can read you can get a book and self study and pass the exams for about $200 per exam (MS exams). If you think that those costs are prohibitive, then you are in the wrong line of work, or are living in the 1940's.