Sign in with
Sign up | Sign in
Your question

What apps will actually use faster RAM?

Last response: in Memory
Share
August 21, 2002 12:06:00 AM

Hi all,
I'm considering upgrading and am pondering the questions of 400vs 533mhz FSbuses and 266DDR ram vs 333.

Obviously the bigger numbers sound better, but they may or may not be worth the extra money, depending on whether my apps will ever use the difference.

Looking at reviews here, and on other sites, most benchmark tests (using current business and or games type tests) seem to show very little difference (yet) between the options when it comes to "real world" computing.

This is not unexpected - as "regular" applications usually don't need the full potential on offer from the latest machines.

But what sort of app WOULD show up a difference? I know that the potential for faster data transfer is there but what would actually need it?

My most demanding apps are games (which I usually run at 1024x768 or 1280x960) so would I see any difference in say load times or any other feature? I already have a reasonably good graphics card and tend to upgrade around once a year.

Would I ever see any real difference, or should I leave 333 ram and 533fsb options until next time around, when there will be more boards around, less chance of incompatibilities, bugs etc.??

Any tips welcome please.

More about : apps faster ram

August 22, 2002 12:27:46 AM

Nobody know?

Perhaps I could try it from a different angle.

I have a GeForce4 Ti card which handles the running of games very well, but I'd like to see if I can get an increase in the time that it takes to load save games (for example).

Which component should I upgrade to get the best improvement, or won't I see much difference past a certain point anyway? And will the results vary with different games that might be loading different kinds of information (or is data just data in these cases?)

Should I go for the fastest CPU, the fastest Front Side Bus speed, the fastest RAM speed, or a bigger amount of RAM?

Which one is most important for loading game saves - or is it another of those eternal "it all depends" computer questions? :-)
a b } Memory
August 22, 2002 3:34:18 AM

It really depends on the platform. AMD chips are using a DDR266 bus, so, DDR333 does very little for them, since the processor bus is limitted to DDR266. But the P4 uses a QDR533 bus, so DDR333 has a significant advantage over DDR266 when you limit yourself to DDR chipsets.

The advantage is mostly seen in games and other high bandwidth applications. You're not going to see a difference in Internet Explorer or Microsoft Word!

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
Related resources
August 22, 2002 3:37:34 AM

Need to know more about your system.
August 22, 2002 7:54:35 AM

I currently have an Athlon 1.2 Gig with 512 megs PC133 RAM and a GeForce4 Ti4200. It runs my games very well (I tend to play at 1024x768 or 1280x960).

But I like to freshen things up about once a year (can't stop scratching that old upgrade itch..!)

I know from experience that the theory and what actually happens in practise are often two different things. So I'd like to learn more before I upgrade this time, and have a clearer idea of the inner workings.

I know that the RAM increase will not show much on screen difference. I have tried running up to date games with both 256 and 512 meg of ram in and have never yet seen a difference on the screen. No doubt if I found just the right spot in a certain game, or used the right benchtesting utility I could find a difference, but it's not a pressing need.

If I go up from PC133 to DDR 266 I will already be getting a fair boost and I can't belive that the extra 25% to 333 would make any difference in practice.

I posted a similar question on another forum and somebody told me that me current specs were already OK for now and that the best improvement I could make would be with SCSI or RAID solutions to speed up hard drive data transfer (to boost the save game load times).

Given that I never leave my system alone for more than 12 months, I'm fairly sure that buying the latest, fastest, meanest solution would be something of a waste of money (satisfying though... )

I'm currently thinking along the lines of something around 2Gig CPU with 512 266DDR ram and some kind of improved hard drive solution.

I currently have a 7200 RPM Seagate Barracuda 20 gig which uses the 80 strand cable (100 ATA???). There is also a D drive at 5400 on the same cable (for backup only), but according to the benchtest utility I ran it doesn't slow down the main drive (which I thought it would - so maybe I should try real world tests by timing loading save games).

Any tips or suggestions about what works in the real world rather than in theory or test suites please?
August 22, 2002 9:11:27 AM

well switching from a SDRAM motherboard to a modern DDR motherboard you will see a difference, allthough your millage will vary depending on the application.

many games tend to be memory bandwidth hungry, especially quake3 and UT.
other particulary hungry apps include MP3 encoding and especially DIVX creation.
when i updated from a KT133A (sdram) to a KT333 (pc2700 DDR) my divx encoding speeds went from 17fps to above 20.
allthough a fair amount of this performance boost depends on the chipset (motherboard). the first DDR boards were quite weak, and rarely got more than 5% performance gains. (KT266). fortunately current ones like the SIS solutions, the via KT266A/KT333 and the Nforce are much better.

as for your harddrives, both on the same cable should be allright, but drive to drive transfers may be slowed down due to the shared bandwidth. having one drive per cable is the ideal scenario.

finally, if you want a super fast IDE hard drive, possibly one of the best buys around is the Western digital 800JB. very fast drive.

<b>MegaHertz Matters! ... But not without Cache our a decent chipset!!! :cool: </b>
a b } Memory
August 22, 2002 9:37:41 AM

You have PC133. I would upgrade the processor, CPU and RAM, all in one shot. PC2100 would get you about 12% better performance on the faster CPU, that's nothing to scoff at!
DDR333, aka PC2700, would only be about 1-3% better than PC2100 but gives you better options for overclocking.

Everything else looks good, I wouldn't improve your drive unless you need more capacity.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 22, 2002 8:17:39 PM

Quote:
I would upgrade the processor, CPU and RAM, all in one shot.

Hmm, I didn't know Processor and CPU are two different devices in a PC? :wink:

--
And now, an advice from your friendly Nike shoes slogan: JUST DO HER!
a b } Memory
August 22, 2002 10:22:13 PM

LOL, I meant Motherboard, CPU, and RAM, sorry.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 23, 2002 1:14:51 PM

This is a bit long winded, but here goes:

REAL WORLD TESTS:

Well, I tried some tests and the results were quite interesting.

I used 4 different game saves – from Medal of Honour, Neverwinter Nights, Morrowind and Gothic. They all use different engines and the way they store and apply the saved information seems reasonably different. The biggest save was from NWN at nearly 20 megs.

The saves from MOH were small, and the times (when reloading on particular level) were too fast to time. So I used a save from the start of a new level, which needed to call up a stack of level info from the hard drive.

In all cases I used a routines that involved reloading several times before timing so that issues like the need to clear the previous data did not distort the results (and in some cases it would have distorted the result by a large margin).

Here’s what I’ve found so far:

DEFRAGGING:

First test was to defrag the drive. This took two and a half hours and made no difference whatever to any of the times (and to be honest – in 20 years of using computers I’ve never yet seen defrag actually fix anything. Scandisk – yes. Defrag – no.)

CHANGING THE CABLE:

According to the HD FAQ here, the old 40 strand cables limit a drive to ATA33 levels of data delivery.

Changing the cable made NO difference whatever to the load times of all 4 games.

CHANGING TO A 5400 rpm DRIVE:

I copied the whole installation from my 7200 ATA lll Seagate to a more modest 5400 rpm Western Digital Protégé drive. This is a budget drive (although both are supposedly ATA 100 compatible). The Seagate’s overall SiSoft Sandra rating was approximately 50% better than the WD drive. In terms of access time on its own Sandra rates the Seagate at 10ms and the WD at 13ms.

This made NO difference whatever to the load times of any of the games.

OVERCLOCKING THE CPU:

Unfortunately I am not an experienced overclocker so I couldn’t get good data here. Simply changing the multiplier from 9 to 9.5 didn’t appear to make any difference, but it’s a very small percentage anyway, so would be hard to pick. I tried 10x but the computer crashed when loading the games. Fiddling with the voltage was probably needed (but I didn’t know to what figures).

CHANGING THE AMOUNT OF RAM:

Now we’re talking! Going back to 128megs (from 512) made a BIG difference. It more than doubled the time for three of the saves to load. The exception was Morrowind, which only went from 15 secs to 19 secs. NWN went from 14 to 34 secs.

Stepping up to 256 dropped all times back significantly. Running two sticks of 128 versus a single one of 256 showed no gains.

Going right up to 512 (albeit in a rather messy configuration of 1x256 + 2x128) made no more difference in 3 of the games, but did continue to improve Medal Of Honour – for which the figures were 9secs(512 RAM), 12 secs (256) and 19 secs (128).

I also tried moving the Windows swap file to the D drive (which apparently helps some programs – e.g. Photoshop) but it made no difference.

So what do you reckon that I’ve learned?

I think that it’s ruled out any need to spend money on faster hard drives, SCSIs or RAIDs as the current drive seem to be able to serve up the data fast enough for these apps (not so for all apps of course). A fresh format and reinstall often makes new drives look faster than they actually are (compared to old, cluttered and buggy ones) so I’m a bit wary of thinking that I’d see a real improvement by buying a new drive.

I’ve got 128 megs of old PC100 ram in another machine, so I might try that and see if that makes it slower again. It looks as if I’ll need to jumper the basic clock speed of the board back to 100mhz and run the CPU at 12x100 to get the 1.2 again. Does that sound right or are Athlon 1.2s supposed to run at 9x133? I’d better check that I guess.

Any opinions or comments on what I think that I’ve learned? Will jumping from a 266mhz FSB to a 400 or 533 help (bearing in mind that I’m also going from an AMD to an Intel, so the jump won’t be as big as it looks)? Will the jump from PC133 to DDR266 RAM boost load speed too (or just boost something else?).

Any other simple tests worth doing?

Place your bets ladies and gents - I’ll let you know the results after I get round to upgrading.
September 15, 2002 9:21:09 AM

Whoa, that was some feat of tedium and patience. Very practical info using real world apps instead of benchmarks.

Your tests seem to confirm the well known fact that very few apps take advantage of maximum bandwidth of even the slowest ata standard and that many drives have yet to reach that same bandwith in actual practice. I think I read somewhere that even the wd1200jb bandwidth was 50mb/sec and furthermore that that performance was considered outstanding, which leads one to believe most other drives are much slower.

Quality is better than name brand, even regarding beloved AMD.
September 16, 2002 2:10:00 PM

Hmmm. Very interesting, and enlightening. Odd about the defrag, I've measured some time difference on boot, but never tested games.Also the cable, check that was a good 80 pin (not doubting ya though). Amazing about the HDD RPM. Try turning up the frontsidebus if you can instead of multiplier this is where the real diff. comes. Overall some very good info. Thanx

:eek:  <b>Who needs heatsinks and fans, I have an igloo</b> :eek: 
September 18, 2002 3:22:06 PM

Dang Ironstone, you should go and work for THG. YOur post was more informative than some of their articles. About the hard drives...When I was a wee lad I had an old FIC MOBO that only supported ATA33. When I put in a Promise Tx2 card it seemed that my computer had doubled its speed. With a WD at 7200 rpm you are already on the bright side of life, but I've heard amazing things about their special edition drives. A 60 GB model is only about 115.00 right now. If your looking for faster load times, this would seem to be the most cost effective way to do it. You could also buy a duplicate of what you have now and put them in raid mode.

Is there a synonym for thesaurus?
!