Cache Makes a Big Difference?

[Yaemish]

I have an overclocked E6300 (running it at 2.34 Ghz with my RAM at 667 4-4-4-12 and running 32 bit Vista).

My new work PC is an E6700 Dell running 32-Bit XP Pro.

Opening programs, word, firefox, and explorer just open instantaneously on the Dell. No lag at all. I thought that Vista's use of memory was supposed to make it appear faster. Am I seeing that the extra cache is making that much of a difference or am I more likely seeing the difference between XP and Vista?

 

[Nenu]

If you are using Aero, turn it off and see if things speed up.

 

[mikeblas]

How are you timing the actions, and what are the specific numbers?

 

[Yaemish]

I haven't timed anything, but there is a definite difference from the user experience alone. I'll try turning off Aero and seeing what happens. I didn't think about it before, it could be the animation that is slowing it down.

 

[Yaemish]

Looks like Aero doesn't make a difference at all. To get a better comparison of the office apps, I need to load 2003 onto my home PC. Also, I just noticed IE opens up in an instant (never noticed this before mainly because I hate IE) but Firfox takes a bit longer. Anyone else notice that?

 

[l3ender]

How much RAM do you have in each?

 

[Yaemish]

At work I have 4 and home I have 2.

 

[RamonGTP]

Hard drive configuration might be playing a pretty big roll here. Any idea on what the Dell is running? Raid 0? Raptors? SCSI/SAS? or a combination of those? If yes to any of those, there is your answer.

 

[mikeblas]

I haven't timed anything, but there is a definite difference from the user experience alone.

Maybe, or maybe not. Until you have hard numbers, you're chasing your tail.

 

[RamonGTP]

Maybe, or maybe not. Until you have hard numbers, you're chasing your tail.

You don't have to physically time something to know there is a difference in performance, you simply have to have used both machines which he has. I'm pretty sure anyone here can tell a performance difference between a system running an 8800GTX and playing Oblivion versus a system with a 6600GT playing Oblivion without having to run benchmarks or have an FPS counter telling them what they already know by simply using the machines.

 

[mikeblas]

One doesn''t need numbers to think there's a difference. To know there's a difference, one must measure.

 

[RamonGTP]

False... Take the example I gave, you KNOW there is a difference. I'll give you another example... Drive a car with 500hp and drive the same car with 150hp. Let's assume you don't know the exact HP numbers, you WILL KNOW which one is faster, period. To know exactly how much of a difference, yes, you will need to measure, but if the difference is great enough, you will KNOW that a difference exists.

 

[mikeblas]

Will I? What if the more powerful car has a crappier suspension, or worse tyres? Or the more powerful car weighs a lot more? Or has a quieter exhaust? Are my involved but indirect perceptions not going to influence my opinion?

What if the 500-horsepower car is a Mustang and I hate Mustangs, and the 150-horsepower car is a BMW, and I love BMWs? Is my predisposition absolutely certain not to cloud my perception?

At all?

For such a large difference, it might be easy to make a guess. Just the same, it's a guess.

But how do you tell the difference between a 155 hp car and a 150 hp car? Since we don't have any numbers from Yaemish, we don't even know what sort of difference he's experiencing.

What if there's another explanation? For your Oblivion example, we don't know if there's a performance difference between the two machines. One might simply be configured incorrectly, or performing a heavy background task, suffering intermittent hard drive failures, throttling for overheating, and so on. The only reason you think there might be a performance difference is that you know the specs of the hypothetical machines you've given.

Engineers measure thing because it's too hard to make good decisions based on subjective comparisons.

Because Yaemish hasn't measured, we're not even sure if he's comparing the same things. After all, XP and Vista use different visual cues at different times to indicate the status of the systems and applications, so it would be an easy mistake.

I hope nobody here is ready to make the leap that cache (er, processor cache? file system cache? disk cache?) is making the perceived change; there are many other variables to consider, and we don't even know the magnitude of the difference we're trying to investigate.

 

[visaris]

False... Take the example I gave, you KNOW there is a difference. I'll give you another example... Drive a car with 500hp and drive the same car with 150hp. Let's assume you don't know the exact HP numbers, you WILL KNOW which one is faster, period.

The problem is that you are using the word "know" in the flexible sense that it is a measure of personal certainty. A measure of how your strong your emotions feel your intuition is in line with reality. This simply doesn't cut it.

"Know" is intended to be used, not in situations of strong personal conviction (such as "I know there is a god"), but in situations where empirical evidence and proper logic prove something to be so.

Will I? What if the more powerful car has a crappier suspension, or worse tyres? Or the more powerful car weighs a lot more? Or has a quieter exhaust? Are my involved but indirect perceptions not going to influence my opinion?

What if the 500-horsepower car is a Mustang and I hate Mustangs, and the 150-horsepower car is a BMW, and I love BMWs? Is my predisposition absolutely certain not to cloud my perception?

At all?

I'm glad to hear you say that nothing is known until measured and proven; I don't hear that enough these days, and I was starting to worry that I was the only one left.

 

[RamonGTP]

Will I? What if the more powerful car has a crappier suspension, or worse tyres? Or the more powerful car weighs a lot more? Or has a quieter exhaust? Are my involved but indirect perceptions not going to influence my opinion?

What if the 500-horsepower car is a Mustang and I hate Mustangs, and the 150-horsepower car is a BMW, and I love BMWs? Is my predisposition absolutely certain not to cloud my perception?

At all?

For such a large difference, it might be easy to make a guess. Just the same, it's a guess.

But how do you tell the difference between a 155 hp car and a 150 hp car? Since we don't have any numbers from Yaemish, we don't even know what sort of difference he's experiencing.

What if there's another explanation? For your Oblivion example, we don't know if there's a performance difference between the two machines. One might simply be configured incorrectly, or performing a heavy background task, suffering intermittent hard drive failures, throttling for overheating, and so on. The only reason you think there might be a performance difference is that you know the specs of the hypothetical machines you've given.

Engineers measure thing because it's too hard to make good decisions based on subjective comparisons.

Because Yaemish hasn't measured, we're not even sure if he's comparing the same things. After all, XP and Vista use different visual cues at different times to indicate the status of the systems and applications, so it would be an easy mistake.

I hope nobody here is ready to make the leap that cache (er, processor cache? file system cache? disk cache?) is making the perceived change; there are many other variables to consider, and we don't even know the magnitude of the difference we're trying to investigate.

What car you like better and how the machines are configured differently are completely irrelevant. I specifically said the same car with the difference being power. 335i vs 328i if BMW is to your likeing, you won't need a timer to tell you which is faster.

Regardless of how one machine is configured, when there is enough of a performance difference, you can tell that there IS a difference... WHY one is faster than the other is another matter all together and that is not what I'm arguing. There are plenty of things you can figure out and know by simple perception. I can tell you without question if I look at a black car and a white car, that there is a very distinct and obvious difference just by looking at it, I don't need to take a sample of the paint and have a chemical analysis done to it to let me know it's different. If I want to know exactly WHAT the difference is then yes, but not to simply know there is one.

The OP has two computers, opened the same program on both and noticed that one of them opens faster than the other. He doesnt' need a timer to tell him that unless he wants to know exactly how much faster one is than the other. Configuration of the machines comes relavent when you want to figure out why.

 

[mikeblas]

335i vs 328i if BMW is to your likeing, you won't need a timer to tell you which is faster.

It would appear that you've completely missed the point of my post.

 

[pxc]

I like Vista, but it's not as snappy as XP on fast hardware, at least for launching programs. The difference you're suspecting is probably more likely due to the software than the hardware. Your Vista system should get faster with commonly used programs if the prefetch works as advertised.

I have XP and Vista dual boot set up and I can directly compare on the same hardware. The differences might be relatively small, but it's still noticeable.

 

[RamonGTP]

How have I missed your point? You drive the two cars, you look at the speedometer while flooring it, and you can tell that one car's speedo is moving up at a significanly faster rate than the other. Same car, different levels of power, enough of a difference that you can see by simple observation, no timers

Two computers, opening the same program, and there is enough of a difference between the two that he can tell one is opening the program faster than the other, no timers.

 

[oozish]

If you are using Aero, turn it off and see if things speed up.

This statement is not true and turning Aero off has been shown to make an insignificant change in VISTA responsiveness.

 

[oozish]

I like Vista, but it's not as snappy as XP on fast hardware, at least for launching programs. The difference you're suspecting is probably more likely due to the software than the hardware. Your Vista system should get faster with commonly used programs if the prefetch works as advertised.

I have XP and Vista dual boot set up and I can directly compare on the same hardware. The differences might be relatively small, but it's still noticeable.

What are you smoking? VISTA launches my programs way faster and general system responsiveness smokes XP with 2-3gigs of RAM. I've gone back and forth between OS's and I prefer VISTA for this very reason.

 

[DemonDiablo]

Will I? What if the more powerful car has a crappier suspension, or worse tyres? Or the more powerful car weighs a lot more?

Stopped reading here. He clearly said the same car. Only person chasing their tail here is you. From the first post you made until now in this thread has been pretty much useless for the OP.

 

[visaris]

How have I missed your point? You drive the two cars, you look at the speedometer while flooring it, and you can tell that one car's speedo is moving up at a significanly faster rate than the other.

So, in this example, one looks at the speedometer, and counts, say in seconds, how long it takes to reach, say 100mi/h, by looking at a clock: 1... 2... 3... This is a measurement, though, one with very poor accuracy and poor precision (yes, these two are different concepts).

This involved two measurements, and I may be willing to accept results from such a test, even though it could have been done in a much better way.

Two computers, opening the same program, and there is enough of a difference between the two that he can tell one is opening the program faster than the other, no timers.

While the first example was a very poor measurement, this one doesn't involve any measurement whatsoever. Worse, this one is based on times, all of which are much less than a second. The car example could at least span multiple seconds.

Your assertion that the two examples are the same or even similar is laughable. You still don't seem to grasp that there is a huge difference between being convinced of something, and proving or knowing something.

At one point in time Americans were certain that blacks were not people. They were treated as big dumb sub-humans. Most people were certain of this savage nature. Did anyone prove this nature? No! This is where emotions, intuition, and superstition come in to fsck-up life for others.

I simply can't stress this enough: One doesn't know something to be true, until one has tested this something with empirical evidence and sound logical proof. Even then, any good scientist will say that nothing is known, only that a theory has been created, which, to date, has been in line with reality. Science and the quest for knowledge is not based on blind faith and accepting things as concrete facts. It is based on constantly adapting our understanding to keep the theory as accurate a reflection of reality as possible.

It would appear that you've completely missed the point of my post.

True.

 

[pxc]

What are you smoking? VISTA launches my programs way faster and general system responsiveness smokes XP with 2-3gigs of RAM. I've gone back and forth between OS's and I prefer VISTA for this very reason.

So have I and Vista feels a bit more poky launching most programs. I have 4GB on my main Vista system, which shows up as 3.2xGB under XP.

 

[mikeblas]

Stopped reading here. He clearly said the same car.

Even if I misread the post, the point remains that you can't compare performance without measurement. Just the same, the example RamonGTP gave was defeated by visaris' valid argument.

From the first post you made until now in this thread has been pretty much useless for the OP.

I'm sure my posts have made him re-think his question. There's little doubt that he's considered that posting more information about the involves systems, thought about clarifying his question, or providing more information about what it is he's been measuring. That's value enough. And if he decides to act on those thoughts, super bonus!!1!

 

[conscript]

think you guys getting a little semantic with the car analogy

It is much simpler to just think in terms of pickign up two boxes. One is extremely heavy, one is light as a feather. You pick each up, and you know which is heavier. You don't need a scale to tell. Sure, if you want to know precisely how much heavier, you'd need to measure with scale, but you don't need to measure to know one is heavier than the other.

Same difference here. Ofcourse he can make the judgement one is faster than the other. Now if you want to say his reality is suspect to not being able to tell the difference...then you're just getting personal

And for this simple post, i submit i should be Hard|OCP poster of the month, september 2007

 

[mikeblas]

What if one box has three matches in it, and the other four?

Yaemish says one machine is at home and the other at work.

Pick up a box of matches at work. Finish the day, drive home. Find the matches at home; pick it up. Hours later, which one is heavier?

 

[swanysto]

I actually thought this was going to be a thread with knowledgeable answers, and not car analogies. The fact is...he can tell a real world difference in load times. Apparently you are trying to prove him wrong, or make him think he is seeing things, unless there is statistical proof behind it. I am sure that is not what he is going for, and pretty sure he hasn't re-thought his question.

To the question at hand, could the ram in the work computer be faster than the ram on his machine? I know dell doesn't usually put the most performance friendly ram in their computers, but that could make a difference. I also agree with the person who suggested HDD's. When I upgraded hard drives I saw a pretty clear gain in response times of the OS.

I started off hating Vista, but once I used it for a while, going back to XP was kind of a drag. Maybe it was just me, but the 64bit version of Vista felt smoother to me than the 32bit version, anyone else feel that way?

 

[mikeblas]

Apparently you are trying to prove him wrong, or make him think he is seeing things, unless there is statistical proof behind it. I am sure that is not what he is going for, and pretty sure he hasn't re-thought his question.

Nope. I'm trying to figure out what's causing the difference he's seeing. He asked:

Am I seeing that the extra cache is making that much of a difference or am I more likely seeing the difference between XP and Vista?

Nobody can propose a sensible answer this question until they get more information:

1. How much difference is he seeing, quantitatively?
2. What is it that he's measuring, specifically?
3. What are the specific differences between the machines?
4. Which cache is he asking about?

To the question at hand, could the ram in the work computer be faster than the ram on his machine? I know dell doesn't usually put the most performance friendly ram in their computers, but that could make a difference.

It might be faster, but any difference in memory speed probably isn't contributing to the difference he's seeing in loading applications. The primary bottleneck in application startup is disk latency. As RamonGTP asked, going to question #3 above, we need to know about the disk configuraiton. And if it turns out he's asking about file system cache, we aslo need to know what test he's using. If he's starting an application he's already run, for example, the application might be starting from the file system cache instead of from paging operations.

Maybe it was just me, but the 64bit version of Vista felt smoother to me than the 32bit version, anyone else feel that way?

If you want knowledgeable answers, why are you looking for feelings?

 

[Comte]

This statement is not true and turning Aero off has been shown to make an insignificant change in VISTA responsiveness.

And your reading into the statement made. He didnt say it did slow things down, but to try it.

 

[joj]

It might be faster, but any difference in memory speed probably isn't contributing to the difference he's seeing in loading applications.

Says who? You?
You just jumped into your own belly >.<
Did you measure the difference in memory speed between computers? You think using the word 'probably' saved you? If you're willing to give real world help without empiric lab testing and measuring why the hell are you getting on everyone's nerves by going crazy about semantics and totally not trying to help OP out in the first place?

If you want measurements and can't help him without it and can't trust his judgment, who the hell are you that we are gonna trust yours? Measure then come back and post.

The primary bottleneck in application startup is disk latency. As RamonGTP asked, going to question #3 above, we need to know about the disk configuraiton. And if it turns out he's asking about file system cache, we aslo need to know what test he's using. If he's starting an application he's already run, for example, the application might be starting from the file system cache instead of from paging operations.

Again, says who, you? Did you measure? How do you know it's disk latency and not for example amount of available memory without measuring? How do you know it isn't some trash/spyware/virus/driver conflict that is making everything appear laggy without measuring? Get real man and take your own advice and don't post before OP "measures" everything out for you.

If you want knowledgeable answers, why are you looking for feelings?

Yeah right, give us a break. People just want quick real world help without going into the problems too deep. If you can give some advice fine, just don't go on goddamn semantics crusade and totally change the direction of the thread into something that helps no one but your ego. Give your best help and let others do the same.


Now to get back on track... Yaemish...

CPU cache is highly unlikely to have anything to do with slower application openings you're experiencing. Vista's use of memory here is irrelevant. XP. because it runs no aero should in most cases seem faster and more responsive.

Slow application startup has a very broad range of reasons. Amout of applications installed on system, defragmentation of the disk, registry problems, hard drives speeds and configuration (raid0, nonraid etc), spyware/viruses/malware thats on machine, antivirus software can have a big impact on apps startups, pure difference in CPU power (which directly influences in speed results from memory and disks), incorrect graphic drivers can create a sort of laggy feeling and so on.

As you can see too many different reasons require more info on your behalf or simply do more research on each of topics i mentioned and compare two pcs.

Of all situations mentioned i would say antivirus apps will surely slow your system down, huge difference in memory timings and speed and amount of apps installed + longetivity of your current windows installation would be what i would focus on in solving your problem. Fresh install of windows will always appear less laggy.

 

[RamonGTP]

So, in this example, one looks at the speedometer, and counts, say in seconds, how long it takes to reach, say 100mi/h, by looking at a clock: 1... 2... 3... This is a measurement, though, one with very poor accuracy and poor precision (yes, these two are different concepts).

This involved two measurements, and I may be willing to accept results from such a test, even though it could have been done in a much better way.


While the first example was a very poor measurement, this one doesn't involve any measurement whatsoever. Worse, this one is based on times, all of which are much less than a second. The car example could at least span multiple seconds.

Your assertion that the two examples are the same or even similar is laughable. You still don't seem to grasp that there is a huge difference between being convinced of something, and proving or knowing something.

At one point in time Americans were certain that blacks were not people. They were treated as big dumb sub-humans. Most people were certain of this savage nature. Did anyone prove this nature? No! This is where emotions, intuition, and superstition come in to fsck-up life for others.

I simply can't stress this enough: One doesn't know something to be true, until one has tested this something with empirical evidence and sound logical proof. Even then, any good scientist will say that nothing is known, only that a theory has been created, which, to date, has been in line with reality. Science and the quest for knowledge is not based on blind faith and accepting things as concrete facts. It is based on constantly adapting our understanding to keep the theory as accurate a reflection of reality as possible.



True.

He doens't need to take a stopwatch with him to work becuase he does not need to know exactly how much that difference is to know that a difference exists.

I'll say it again for like what... the 4th time? When the difference is great enough, an exact measurement to see that something is faster than another. If you're comparing two very similar things, then yes, accurate measure would be needed

He can tell there is a differene because a large enough difference exists, period.

.

 

[mikeblas]

Again, says who, you? Did you measure?

Yep. I've been writing and tuning Windows applications for about 15 years. If you don't believe me, measure it yourself -- to tools to do so are free and easy to find. If you don't know how to measure it yourself, get a book like Windows Internals and read up on how the loader works.

How do you know it's disk latency and not for example amount of available memory without measuring?

Like I said, I have measured I've been measuring for a long time, and I'll measure again and again, very soon.

How do you know it isn't some trash/spyware/virus/driver conflict that is making everything appear laggy without measuring? Get real man and take your own advice and don't post before OP "measures" everything out for you.

Sorry; I thought it was obvious that I was offering general descriptions of the relationship between the effect of the different subsystem on application loading and not prescriptive advice for Yaemish's specific case. Since you've repeatedly pointed out the questions I've asked Yaemish about his circumstances and tests, I would've thought you were well-aware of the additional information needed to make a specific diagnosis.

Your attempt to "get back on track" demonstrates as much.

Vista's use of memory here is irrelevant. XP. because it runs no aero should in most cases seem faster and more responsive.

Vista is more aggressive about keeping files in cache than XP. As such, Vista's use of memory might be very much relevant. If Yaemish has previously run the application he's observing, it's likely that the file system cache is charged with the data needed to load and run the app and lots less I/O needs to happen. Or, if he's running different applications as he tests, the file cache is missing and disk subsystem throughput is more important.

That is, either you know something Yaemish hasn't told us (how he's testing and what he's testing), or you have some insight which invalidates the common knowledge that I/O is the slowest part of running an app.

Which is it?

XP. because it runs no aero should in most cases seem faster and more responsive.

Did I miss where Yaemish mentions he's running Aero? You do know it can be disabled, right?

Slow application startup has a very broad range of reasons. Amout of applications installed on system,

How does the number of applications installed on a system influence the startup time of one particular application? Tell us: is the relationship linear, or exponential?

The problem with your method of "getting back on track" is that you're going off half-cocked, using flawed opinions of how things work and an incomplete picture of the problem scenario to try to give advice. This only confuses the person with the problem: now, they're worse off, and they're believing your half-baked explanation.

 

[engiNURD]

K, I got lost amidst the arguing about cars n all, so I'll just toss this in there... I'm not sure if its already been said or not, though, so I'm sorry if it has...

How long have you been running Vista? It takes some time to learn your habits before it speeds up from its learning process.

The cache difference between 2 and 4MB shouldn't be that much of a noticeable performance difference. I have an E6400 at home and an E6700 at work, both running XP 32 bit, and my home system "seems" faster than my E6700... probably because its OC'd to 3.6Ghz, but still... lol.

 

[Yaemish]

Wow, this thread went to hell....

I ran sys info at work, so now I have an idea of what is installed..

Work
e6700
4 GB of 666 mhz memory
Seagate Barracuda Perpendicular drive (SATA2 but not sure if it is in SATA2 mode)
Windows XP 32 Bit
McAfee 8.5

Home
e6300 (clocked to 2.4 ghz)
2 GB of 666 mhz memory (4-4-4-12)
2 Seagate Barracuda Perpendicular drives (SATA2 with the applications installed on second drive)
Vista 32-BIt
McAfee 8.5

I need to install Office 2003 at home so I have a better comparison (2007 is slower). I was using Firefox and how quickly it opens as a point of comparison... Not sure if that was a good idea because I think that Firefox may be slower under Vista. Opening IE is lightning fast on both systems, Firefox is slow on the home system.

 

[joj]

IE will always open fast because thats how microsoft integrated it so its practically open all the time. Now firefox slowdown is different. If you have lots of addons installed firefox startup will slow down dramatically. When i say lots of i mean even 6-7. Try without addons if you have any. If not then it's something else.

Check your processes running and compare to pc at work.

 

[joj]

Yep. I've been writing and tuning Windows applications for about 15 years. If you don't believe me, measure it yourself

10 LAWL!
Did you measure his pc? No? Surely you must know that various computers give various performace? To be more scientific we must take into account different behaviors of different computers and since input variables here are clearly different (probability that you have exact system as him is almost zero) you shouldn't be doing such comparisons when giving advices.Take you're own advice and don't post until he measures it out for you Mr. Science.

Like I said, I have measured I've been measuring for a long time, and I'll measure again and again, very soon.

20 GOTO 10

Sorry; I thought it was obvious that I was offering general descriptions of the relationship between the effect of the different subsystem on application loading and not prescriptive advice for Yaemish's specific case.

O'RLY? All you did was going on egoistic semantic crusade without helping anyone and now you're like giving advices from your experience that you should have done in the first place and save us all that crap at the beginning of thread.

Vista is more aggressive about keeping files in cache than XP. As such, Vista's use of memory might be very much relevant.

If my pc works without such problems that Yaemish has and i have dual boot xp sp2 + vista then surely vista's memory use isn't in question here. Specially if he has enough memory which he has. Don't generally blame OS if it works without such problems for other people. So no, Vista's use of memory shouldn't have much effect here.

The problem with your method of "getting back on track" is that you're going off half-cocked, using flawed opinions of how things work and an incomplete picture of the problem scenario to try to give advice. This only confuses the person with the problem: now, they're worse off, and they're believing your half-baked explanation.

LAWL! They are just splendid with advices you gave in this thread... oh dear, i hear someone's ego growing as he prepares to type reply... I won't be replying to you no more so bye bye

 

[RacerX27]

What was the question again?

OH ya, I remember.

New PC at work.... my first thought is try a defrag on the home system. Nothing like the fresh fast feel of a new install. A defrag is a close second to that fast feeling.

Worth a try..... if that doesn't work drop a Hemi in it.

 

[gwai lo]

10 LAWL!
Did you measure his pc? No? Surely you must know that various computers give various performace? To be more scientific we must take into account different behaviors of different computers and since input variables here are clearly different (probability that you have exact system as him is almost zero) you shouldn't be doing such comparisons when giving advices.Take you're own advice and don't post until he measures it out for you Mr. Science.

20 GOTO 10

O'RLY? All you did was going on egoistic semantic crusade without helping anyone and now you're like giving advices from your experience that you should have done in the first place and save us all that crap at the beginning of thread.

Dude, you totally missed the point. It's not about all this crap that you guys are bringing up, it was all about measuring and "feeling". Chill out and go back and reread it all.

 

[Arcygenical]

Holy Crap. Someone took a huge dump on this thread.

OP: I noticed the same thing with Vista too, thus I reverted back to XP and never looked back (forward?).

 

[engiNURD]

Holy Crap. Someone took a huge dump on this thread.

OP: I noticed the same thing with Vista too, thus I reverted back to XP and never looked back (forward?).

How long did you use Vista for? As I stated earlier, it takes a while for Vista to learn your tendencies. Once the initial learning process is finished, it speeds up noticeably.