How do I speed up my PC or what should I replace?

1

I have the following config:

  • Core 2 Duo E7400 @ 2.8 GHz
  • 1 GB DDR RAM 667 MHz
  • 250 GB HDD @ 5400 RPM
  • Intel DG31PR Motherboard-

This is an April 2009 Computer and running Windows 7 32 bit. The performance on loading 3-4 programs almost hangs the computer and soetimes the mouse pointer also lags. And one more thing is it shows a continuous high Hard-Disk usage.
What may be the problematic part and which part is causing it?
Thanks in advance.

Arumoy Chakraborty

Posted 2014-08-10T10:58:38.950

Reputation: 45

1It may be your HDD limiting cause of memory swapping to disk (cause 1GB is quite low for Windows 7, my 32-bit machines usually are taking 1-2 GB without anything running) but it could also be any other problem (the lagging mouse pointer seems strange), especially faulty drivers – th3falc0n – 2014-08-10T12:37:28.063

On the short term, I'd suggest getting a small USB drive and experimenting with readyboost. It helps in some scenarios, and may take some of the load off your system. I'd probably do this in addition to the hardware upgrades suggested. I'd add that getting parts for older system is a pain unless you can find somewhere that sells refurbs or pulled parts, and you may have to do that for a ram upgrade. Memtest may be your friend here if you end up with a dud stick – Journeyman Geek – 2014-08-10T13:01:58.730

Buy SSD and forget about all that "readyboost" nonsense forever. If you can't afford to get a big one, look for a 32-40 GB model. It should be relatively cheap and will fit WinOS and a few essential apps, like MS Office and whatnot. – user1306322 – 2014-08-10T13:22:12.240

Note that the DG31PR uses DDR2, not DDR. Obtaining DDR2 RAM should not pose a problem.

– a CVn – 2014-08-10T13:46:25.823

Answers

4

Like Carles Fenoy said, start with the RAM. 1 GB is the minimum required for 32-bit Windows 7, which generally means something like "the system will boot, and you will be able to start Notepad, but expect to get plenty of coffee while you wait for it to open". It also doesn't take into account the requirements of other applications that you may want to use to actually get useful work done.

Upgrade your RAM to at least 2 GB, preferably 3 GB. (Note that you may need to alter the boot parameters in order to have Windows make use of the range 2-3 GB.) Don't go above 4 GB on a 32-bit system; Windows won't use the additional RAM, so you'll effectively be wasting your money by installing more than 4 GB. The DG31PR sports two DIMM slots, which means that you should be able to move to 3 GB by installing a second 2 GB DIMM; moving to 4 GB would require replacing the existing DIMM. Note that it appears to use DDR2 RAM, not DDR.

After that, a 5400 rpm hard disk is not very fast by any measure. I would replace it by at least a 7200 rpm model, and would seriously consider a SSD which would greatly help with I/O latency, which is most likely the limiting factor in your disk I/O. (Use the system performance monitor to look at disk I/O queue length; that'll tell you if the disk is a bottleneck. Higher numbers are worse.) Get an overprovisioned model, put a system-managed swap file on it, and accept that it might wear out slightly sooner than a spinning-platter hard disk drive perhaps would. From a flash wear perspective, it'll last long enough to serve a useful life; you're more likely to see e.g. controller failure than flash wearout with reasonable use. You can get an external enclosure and relegate the current drive to serve as backup storage if you wish (you can never have too many backups). A SSD should also cut the power consumption of the computer by somewhere around 5-10 W compared to a spinning-platter hard disk, as well as lower the noise level of the computer because there are fewer moving parts.

With these two changes (upgrading to a total of 3 GB RAM, and a SSD) you'll probably find the performance of that system to be quite acceptable.

a CVn

Posted 2014-08-10T10:58:38.950

Reputation: 26 553

1Considering the age of the system, and endurance on modern drives, I suspect the SSD may outlive the rest of the system. – Journeyman Geek – 2014-08-10T12:50:56.773

@JourneymanGeek Certainly possible, but my previous system served for the better part of a decade (2004 to 2012, IIRC; incremental upgrades, but some parts remained the same throughout that time). I mostly wanted to point out that flash wear is not likely to be what eventually kills the SSD. – a CVn – 2014-08-10T12:54:04.127

Platter density makes a bigger difference than whether it's 5400 vs 7200 RPM, which means a 5400RPM HD could be as fast or faster than a 7200. – vol7ron – 2014-08-10T14:37:59.857

2

From my point of view your configuration seems to be very short on RAM. The high HDD usage is because of the use of the disk as paging file, so when there is not enough memory available some is freed and moved to the HDD. So increasing the amount of RAM should considerably increase the performance.

Carles Fenoy

Posted 2014-08-10T10:58:38.950

Reputation: 151

Note that 32-bit Windows can support up to ~3 GB of RAM, which should already greatly improve performance. – gronostaj – 2014-08-10T12:35:21.670

1

First, I will say what you're running and what you're opening could have an extreme impact on the perceived time to open applications and will also have impact on what you need to upgrade, or if you need to upgrade a component. Ctrl + shift + Esc, and open the resource monitor and keep it open while opening those applications to see what the memory usage is like - it is most likely the culprit.


Biggest Bet

In general, based on your specs, upgrade:

  1. RAM

I'd suggest 4GB for your system. Unlike what others have said, 4GB can be used in a 32b system. If you want to test out a 64b system for free you could download one of the Linux distros (e.g., Ubuntu).


Why not HD?

Unlike what other users have said, you do not need to upgrade your HD. Granted a 2009 5400RPM HDD is a little dated, but sometime around then many 5400RPMs were performing the same or better than 7200RPMs, so it is a matter of manufacturer. I've "invested" thousands in SSDs for many many years now and yes, while they will speed up application loads and boot times, you are experiencing a lag that will exist if you get one. Also, SSDs are still expensive than getting another stick of RAM.


What else could it be?

For 1, if you're running antivirus or other background applications/service, they could certainly be consuming memory (and CPU cycles) that you may desperately need. Furthermore, applications like Photoshop and video games may need a considerable GPU with a decent proc and GDDR, so yes, upgrading your video card may also help performance.

Also, over time, parts go bad, especially when they've been exposed to extreme heat or cold. That means that the RAM you have could have bad blocks, as well as the HD. So there is the possibility that you might need to upgrade one of the other components in addition to giving it some more temp memory (RAM).

Lastly, one thing you should do is open up that case and make sure to use some compressed air to blow out dust. Dust is bad for the computer as it could bridge paths on the motherboard that should not be connected. Due to failsafes, many times the computer will still operate, but slower, eventually leading to overheating and a crash. Regardless, your computer is an engine, and like a car it should be maintained.

vol7ron

Posted 2014-08-10T10:58:38.950

Reputation: 403

1"over time certain addresses will fail" RAM doesn't work the same way as a storage device, which will (eventually, hopefully) relocate problematic blocks to spare locations (often this happens only after data is already lost). Faulty RAM will simply mean that you'll see corruption, or crashes, and won't be able to determine the cause of those because by the time you investigate, the cause will likely be gone. And I'll bet you a dollar (any country's dollar) that looking at the I/O queue length when the system is under stress will show that storage (swap) is a problem. – a CVn – 2014-08-10T13:54:59.650

Yes, but when it builds the address table, it will not use those faulty blocks in the future. Unless my understanding was incorrect, or somethings changed. I thought it would be excluded from the page table – vol7ron – 2014-08-10T13:56:25.777

"the processor you have may not be as fully functional" - I doubt this is true too. It either works or crashes. CPU's can't work most of the time with exceptions. Related: Is the performance of a CPU affected as it ages?

– gronostaj – 2014-08-10T13:57:44.213

1"when it builds the address table, it will not use those faulty blocks in the future" Citation needed. Note that RAM corruption can be caused both before, during, and after a value is stored at a location. And even if this was true, those tables could just as easily be corrupted as well due to RAM problems. – a CVn – 2014-08-10T13:57:49.877

Interesting how this is getting downvoted as it has not only addressed that an SSD is not *not* necessary, but has added alternative problem cases that may appear. Citation is not needed, I've built and run diagnostics on computers and this isn't Wikipedia. – vol7ron – 2014-08-10T14:00:08.837

@MichaelKjörling agreed corruption is tricky but you're making a mountain out of a molehill for this answer. – vol7ron – 2014-08-10T14:01:49.377

@gronostaj disagree with that answer, it does not talk about all the other components that make up modern CPUs for instance a huge part would be the L1,L2,L3 caches, where a CPU would still be functional even as a part degrades. – vol7ron – 2014-08-10T14:04:32.117

1Actually, in this case, I'd say citation is needed. If you make a statement of fact, you should be prepared to back it up with references, especially if it goes against reasonably expected behavior. I could say on SO "the .NET jitter is written in assembly language by extraterrestrials", make the same claim as you do (basically "I've used it and this isn't Wikipedia") and it'd still very likely be shot down. – a CVn – 2014-08-10T14:06:49.053

@vol7ron Have you seen a single case of a CPU becoming slower due to cache corruption? It's not a HDD, corrupted parts won't be reallocated, it will just fail when using them - either by giving incorrect computation results or by failing completely. Neither of those is what I call functional. – gronostaj – 2014-08-10T14:09:47.850

@MichaelKjörling If you threw that in your answer to a question about Python, the meaning of your answer would not change as it is not relevant - one does not impact that other. However, I didn't expect something to be of such debate, so I will remove it here – vol7ron – 2014-08-10T14:10:33.550

@gronostaj when bad blocks are encountered I believe they are removed from the paging table. I may be under false belief and I have no facts to support this claim as you guys have mentioned, so I am removing it from the answer – vol7ron – 2014-08-10T14:11:49.423

2"when bad blocks are encountered I believe they are removed from the paging table" With non-ECC RAM (which something like 99.999% of non-server systems use), there is no way to know whether you're retrieving a correct or incorrect value from RAM. HDDs dedicate gobs of room to FEC data to guard against corruption and be able to repair it when it happens. (Look at the SMART raw read error count attribute some time. You might be surprised.) Non-ECC RAM has no similar provisions. The only way to reliably detect memory problems is by using ECC, which very few home computers do these days. – a CVn – 2014-08-10T14:15:20.847

@MichaelKjörling Well I've primarily used ECC in my builds, but I don't have what I'd call average devices. That is, except for some of my OC'd boxes, where ECC (at the time) was actually a limitation - not sure how things are today, as it's been many years. – vol7ron – 2014-08-10T14:18:44.647

The DG31PR does not support ECC RAM. So we can safely assume that the OP is not using ECC RAM currently, or at the very least if the DIMM is ECC then the ECC capability is not being used, and it is not practical to move to the RAM having ECC capability without completely rebuilding the system. Hence, anything centered around the RAM's ability to detect corruption is a no-go right from the beginning. – a CVn – 2014-08-10T14:22:32.960

@MichaelKjörling yes I think your 99% figure was probably accurate :) Anyhow, I think I've pulled out the questionable items, which I didn't expect to be so controversial - from my experience (running different configurations, OCing, and marking statistics over the years) forgive me that I still have some notion that it's not just all-or-nothing when it comes to CPUs degradation; but I've removed it as lack of empirical evidence. Thank you all for pointing this out. – vol7ron – 2014-08-10T14:27:16.857

-1

After putting in more and faster RAM (2x 1 GB 800 MHz PC2-6400 RAM, enabling the dual memory architecture of you MoBo), consider replacing your HDD. A 5400 rpm HDD is quite slow. Consider a 7200 rpm SATA disk or even better a SSD disk. Use the SSD for Windows and Applications, while retaining your old HD for data/documents/media.

agtoever

Posted 2014-08-10T10:58:38.950

Reputation: 5 490

32-bit edition of Windows won't support 4 GB of RAM. – gronostaj – 2014-08-10T12:37:27.620

Agree. Changed it to 2x 1 GB. – agtoever – 2014-08-10T12:41:05.323

1I'd rather go for additional 2 GB stick. Additional 1 GB should compensate lack of dual channel, plus there's no need to replace perfectly good 1 GB stick with two (a bit faster) 1 GB sticks. – gronostaj – 2014-08-10T12:43:34.607

RAM latency is virtually never a noticable bottleneck in anything resembling a modern system. It might be under some specialized workloads, but for general office-type use, I can't see it matter. My system is 32 GB at DDR3-1333 (same memory clocking as the OP) and I've never felt RAM latency to be a performance-limiting factor. – a CVn – 2014-08-10T13:07:36.177

3@gronostaj That is WRONG. 32 bit Windows 7 can out of the box utilize up to 4 GB of RAM (and with PAE extensions even more). Some of the 4 GB is may not be addressable because of overlap with memory space required by video-card and other hardware. That still leaves you with minimally 3 GB usable (in most cases as much as 3.75 GB). – Tonny – 2014-08-10T13:16:14.830

-1

  1. Upgrade your RAM.

  2. Go through the task manager tasks and check which is keeping your hard drive busy

  3. Open a CMD and enter: powercfg -getActiveScheme. Copy the GUID.

  4. Open a CMD and enter: powercfg -setacvalueindex [GUID without the square brackets] 54533251-82be-4824-96c1-47b60b740d00 893dee8e-2bef-41e0-89c6-b55d0929964c 100
    and
    powercfg -setdcvalueindex [GUID without the square brackets] 54533251-82be-4824-96c1-47b60b740d00 893dee8e-2bef-41e0-89c6-b55d0929964c 100. This will decrease battery life but increase performance

DividedByZero

Posted 2014-08-10T10:58:38.950

Reputation: 280

1What this does is set the minimum processor clock rate to 100% of full speed. This does nothing to solve the OP's problem, and the OP is not running a laptop. – bwDraco – 2014-08-11T02:20:23.493

It's also totally opaque. If you want to suggest setting the CPU power to always be at 100%, it's much better to tell them to go through Control Panel -> Power Options -> Edit plan settings -> Change advanced power settings -> Processor power management -> Minimum/Maximum processor state. That way, one doesn't need to worry about what the numbers mean. – a CVn – 2014-08-11T07:34:18.117

At the very least, with things like these, you should include an official reference for the relevant GUIDs so that they can easily be verified. Googling for these finds PowerEnumerate (official documentation) for the group GUID but nothing clearly official for the specific GUID (the 893dee8e one), which means that what this does can change at any time since Microsoft are not bound to maintain those specific GUIDs for those specific values. Only official documentation is an interface contract.

– a CVn – 2014-08-11T07:35:51.567

I'm only going off an E6600 I have in a workbench somewhere, but I thought that series proc didn't have the turbo. I thought there wasn't a power up/down, only a dual-core proc. – vol7ron – 2014-08-11T13:22:49.123