Need help with CPU qualities....

Factual Questions
1

Ok, I need some ‘splainin’ about CPU’s or whatever their official designation is.

“Processors?”

My computer has:

600 megahertz AMD Athlon (slot A )(slot 1)??
128 kilobyte primary memory cache
512 kilobyte secondary memory cache

And LadyMack’s computer has:

850 megahertz AMD Duron (slot A )(slot 1)??
128 kilobyte primary memory cache
64 kilobyte secondary memory cache

Okay, as I understand it, LadyMack’s will work a bit faster.

What is the 128 primary for?

What is the secondary for?

Is my bigger secondary why I can defrag and mirror drive to drive 200 to 300 % faster than her machine will?

Or is that more a factor of BIOS?

We have the same 100 Mz FSB. Same HD’s

512 Meg of SDRAM PC133 ( 2 X 256 ) I actually have (2 x 256 + a 128) in mine.

Same Win98se with same updates etc. Mine is more organized as to files and such but she has less stuff. Both are under 25% HD usage.

So, are those cache’s in the processors or on the MB? In the BIOS? Do they account for the speed differences?

Can someone or multiple someone’s ‘splain’ me about this?

Thanks in advance.

3½¢

2

The primary and secondary cache is directly on board the CPU chip itself.
Both store instructions (operating systems code or program code). The CPU will fetch instructions from the primary first, if not available then from the secondary cache. If not available there it looks to your RAM memory for stored instructions. And if not there the CPU will look to a storage medium, such as a hard drive, etc.

The advantage with caches on the CPU die itself is faster response times since electrically, it is a shorter path to the instructions.
Fetching instructions from RAM memory takes longer. (Remember we are talking nano-seconds here; fetching instructions from the hard disk would be measured in micro-seconds).

In short, yes, your larger secondary cache is a large factor in the performance of things like defraging a hard drive, indexing the file system, and other low level functions. You also have slightly more RAM memory to work with, which always helps things along.

I’d still take a 800 Athalon over a 600 any day.
I hope the above explanation didn’t muddy the waters too much.

3

Actually your athalon will probaly perform the same or better than your SO’s Duron in most cases. Architecture and cache size will play a huge role in the final performance of a processor.

4

No, that’s the instruction cache. The L1 and L2 caches are data caches. Basically, memory holds nothing but numbers, and everything a CPU does is really just some kind of arithmetic. The trouble with today’s computers is that a modern CPU could do addition 100s of times in the amount of time it takes to get one number out of memory. This means that the CPU must sit and waste a lot of time when it needs to access memory, which happens a lot. The good news is that programs tend to use the same memory locations over and over again rather than just accessing memory completely randomly. The caches are small but very fast memories that hold the value of recently-used memory locations. The idea is that the CPU always looks in its cache when it tries to access memory. If the memory location is being held in cache, the CPU doesn’t have to wait for the main memory to fetch the value needed.

Think about a secretary who is reading and making changes to a whole bunch of paper records. The filing cabinet is the main memory. His desk is the cache. Rather than going to the always going to the filing cabinet whenever he needs a record, he’ll probably keep a small pile of the records that he’s working on on his desk. This will save him a lot of wasted time looking up things in the filing cabinet.

It has very little to do with their being a shorter path to the cache than the main memory. We can construct memories using different materials and methods. It’s possible to make very fast memories, but they will be prohibitively expensive. It’s also possible to make very cheap memories, but they will be prohibitively slow. Caching is a compromise that lets programs enjoy the speed of the very fast memory much of the time, while only using small amounts of that fast memory so that the total cost isn’t too high.

I have my doubts about this. Because hard drives are so incredibly slow relative to a CPU(and any type of memory), access times to the hard drive should dominate the run-time.

5

But ANY instruction is data. I don’t disagree with your point. Perhaps I was trying to over-simplify for the OP

Excellent analogy

Point noted.

6

Thanks guys, I am going to do a test in a few that might help.

I just re- defragged her back-up hard drive in my machine and it changed the last 10% completely from the way her machine had it.

Now, I’m going to clean her main drive up and then scan and defrag it in my machine then mirror in my machine then put it all back in her machine and re-defrag the main drive and see it it changes the last 10% again.

I can then note the speed of the defrags and the speed of the mirror in my machine with her HD’s. That can help find the low points.

As it stands now, with identical hard drives, her 8 Gigs of data takes 2 hrs to defrag and my 14 Gigs takes 30 minutes if they are both in about the same shape.

7

Totally off topic, but damn man, time to upgrade. Win 98SE??? Ouch.

8

If it does everything they want/need it to do, there is no reason at all to upgrade.
If you have a stable system you should only upgrade when needed not just because there are newer products (software or hardware) available.

9

The L2 cache holds both instructions and data (as one poster noted, instructions are just data) but there may be separate instruction and data L1 caches that fill from the L2 cache (see here ). BIOS has nothing to do with anything once the OS is up and running (at least since DOS days).

I think it will be interesting to see what happens when you put her drive in your machine. Perhaps the file systems are formatted differently. Keep us informed.

10

Yes, as long as I am out in the sticks and on a satellite connection, am not a gamer, do not do online interactive games, just read message boards, post pictures for friends and family and surf the web, why should I upgrade to a machine that just waits around for content to get down the pipe, heck, I have to wait on some places now, a faster machine will not speed me up any. My printer, scanner, network, camera interface, 5 in 1 reader. All my equipment works.

We live on social security so $$ are tight but I can have 5 computers (one dicated Linux machine, no need of ‘dual boot’ )with separate APC backup for the main ones in the house. The laptop has it’s own battery and the one in the garage is for checking out free parts I come to collect.

I have less invested than a single mid-level game machine with modern scan, print super dupper video card, etc.

To each his own.

Back to my speed and CPU question.

Okay, this is really strange.

In my machine right now, I have a 80 Gig WD (7200 RPM) hard drive partitioned to 40 Gigs as my main with my CD burner as slave. On IDE ‘the second’ I have a ‘CD player’ as slave and the hard drive as master.

This works fine with WD, Samsung, Maxtor and Seagate, IBM and a few strange little HD’s I have laying around or with stuff stored on them.

It will fuss about any of my Linux hard drives, it knows they are there is about all. This all seems normal to me.

Now, My Seagate (120Gig) { partitioned to 40 Gigs} (7200 RPM) and 80 Gig WD ( resized to 40 for speed)(7200RPM) are the two I use for main drives and mirror drives. (about a weekly schedule of mirror and swapping) and all that works fine.

The SO’s main WD 20 Gig (5400 RPM) {This hard drive works good in my 'IDE the second} and 120 Gig Seagate not resized, just a full 120 Gig (7200 RPM) { This Seagate will not even show in my ‘IDE the second’ }‘Windows Explorer’ or ‘Belarc’, but does on the boot up. It get it’s lights showing power etc.

My mirror utility, (on a floppy which boots during a cold start ) can see and work with it. :confused:

But I can’t see it to defrag or scan the way I normally do it, which is the only way to compare ‘apples’ to ‘apples’. Actual defragging and scanning is not the problem remember, it is the speed of operation between the two machines that is puzzling me and the cause of all the questions. What is the deal with her machine?

That different BIOS,
The different L2 cache,
A bad Mother Board???

My little 400 Mz CPU with 300+ RAM out in the garage defrags and scans and mirrors faster than the SO’s… ???

It (the garage one) does have the same BIOS I have on my machine… ???

My machine has seen 120 Gig HD’s before just fine. … ???

I normally keep whichever is the back up hard drive for that week physically disconnected from the machine.

I will often have a little 10 Gig or less on my ‘IDE the second’ if I feel the need to have something there or some purpose.

Anyone have any ideas??

Here is the 'right now ‘Belarc’ readouts.

Here is SOs’:

Windows 98 SE (build 4.10.2222)

VIA Technologies, Inc. VT8363

Processor a

850 megahertz AMD Duron

128 kilobyte primary memory cache

64 kilobyte secondary memory cache

Main Circuit Board
Board: 8363-686A

Bus Clock: 100 megahertz

BIOS: Award Software International, Inc. 6.00 PG 11/29/2000

Drives

120.00 Gigabytes Usable Hard Drive Capacity

111.30 Gigabytes Hard Drive Free Space
Memory Modules c,d
CD-ROM Drive/F5E

Generic floppy disk drive (3.5")

ST3120026A [Hard drive] (120.03 GB) – drive 0, s/n 5JT5LFZJ, rev 8.01, SMART Status: Healthy 512 Megabytes Installed Memory
Memory Modules

Slot ‘BANK_0’ has 256 MB

Slot ‘BANK_1’ has 256 MB

Slot ‘BANK_2’ is Empty

Slot ‘BANK_3’ is Empty

Local Drive Volumes

c: (on drive 0) 120.00 GB 111.30 GB free

Here is Mine:

Operating System System Model

Windows 98 SE (build 4.10.2222) No details available

Processor a

600 megahertz AMD Athlon

128 kilobyte primary memory cache

512 kilobyte secondary memory cache

Main board: AMD-75X-W977

Bus Clock: 100 megahertz

BIOS: Award Software International, Inc. 4.51 PG 10/01/99

Drives

80.02 Gigabytes Usable Hard Drive Capacity

65.01 Gigabytes Hard Drive Free Space

ATAPI-CD ROM DRIVE-50MAX [CD-ROM drive]

YAMAHA CRW8424E [CD-ROM drive]

Generic floppy disk drive (3.5")

Generic floppy disk drive (3.5")

Generic IDE hard disk drive (40.01 GB) – drive 1

Generic STORAGE DEVICE [Hard drive] – drive 255

WDC WD400LB-00DNA0 (40.02 GB) [Hard drive] – drive 0 640 Megabytes Installed Memory
Memory Modules

Slot ‘0’ has 256 MB

Slot ‘1’ has 256 MB

Slot ‘2’ has 128 MB

Local Drive Volumes

c: (on drive 0) 40.01 GB 25.01 GB free

d: (on drive 1) 40.01 GB 40.00 GB free

11

Hehe I used a similar analogy when explaning computers to people sometimes. Except I said L1 cache is like stickies on your monitor. Real quick to glance at, but it won’t hold very much at all. L2 cache is the desktop. Takes a little longer to fiddle with and get the info, but still pretty damn fast, and pretty limited. RAM is like the desk drawers. They hold a fair amount, but you got more to look through, and you can get to them pretty quickly. Hard drive is like the big archive in the basement. You have to get up and go down the stairs to get it. Takes much more time, but it holds vast quantities of data.

12

GusNSpot, have you checked the DMA settings in two PCs, both for Win98 (under Control Panel -> System -> Device Manager) and in the BIOSes? All disks – including CD drives – should be set to “DMA” (or “UltraDMA”) mode, not “PIO” mode. Incorrect DMA settings could easily account for a factor of three or four in disk I/O speed between the two machines. Note that if she has an old CD-ROM drive that can’t do DMA, any hard disk on the same cable might be forced to the slower PIO mode when in her PC.

I’d would expect “the SO’s main WD 20 Gig (5400 RPM)” disk to be noticeably slower than the various 7200rpm drives, but that’ll be true whichever PC it’s in.

This is the crucial point. Disk-intensive processes such as file-copying should be almost entirely determined by the disk parameters and interface, together with the relevant BIOS and OS settings.

Caches:
L1 and L2 data cache memory affects speed when the same data blocks need to be accessed repeatedly. For applications that mainly move data onto and off of the disk (such as disk defragmentation), the only repeatedly-used data will be the FAT (File Allocation Table). In both of GusNSpot’s systems, there’s enough main RAM (768MB and 512MB) to hold the necessary copies of the FATs.

Let’s estimate the I/O bandwidths of GusNSpot’s PCs, CPU <-> RAM and CPU <-> disk:
[ul]
[li]RAM: given a 100MHz FSB (Front-side bus) and PC133 RAM (which has a 64-bit data bus), the burst read or write rate between CPU and RAM will be 6400Mbps (=800MBytes/sec). If the motherboards’ memory managers are set up for bank interleaving, it’ll be twice this, but let’s be conservative and say they aren’t.[/li] [li]Disk: almost all modern IDE/ATA drives have sustained I/O bandwidths of less than 40MBytes/sec, no matter whether the interface is ATA66, ATA100, or ATA133. This’ll be slower still on a 5400rpm drive.[/ul] [/li]So, I’d estimate that even GusNSpot’s PC133 RAM is at least 20 times faster than his disk I/O, and that’s not even accounting for drive latency. The L1 and L2 cache are, of course, significantly faster than PC133 RAM.

What the above tells us is that for disk-intensive processes, the L1 and L2 caches play no significant role, and thus differences between their sizes and speeds are irrelevant. If you want to bet on a 100m footrace between an 80-year-old man (=the disk) and an in-form recent Olympic Games 100m medal-winner, you don’t really need to know whether the latter won Gold (=L1 cache), Silver (=L2 cache) or Bronze (=PC133 RAM). :slight_smile:

[Finally, I’m all in favor of keeping Windows 98SE on systems that don’t need anything more (although I’d personally hate not having Google Earth, however slowly!). I actually have a bunch of people for whom I do informal tech support who are on Win98SE, with dual-partitioned HDs containing a Ghost image of the latest fully-stable configuration. Since Win98 can boot to “real” DOS, it’s a 5-minute job to resurrect a PC made sluggish or unstable by spyware, kids downloading games, etc. I teach them how to backup their data – important in any case – and replace the current buggy OS with a clean Ghost image, and they’re good to go for years, since all they generally want to do is the same as GusNSpot and his SO.]

13

Oh Man, great post Antonius Block and armed with it in front of me on this ‘Ubuntu’ puter screen, I’m going to play in the BIOS of the other two.

[que ‘evil monster gonna get you’ music]

If her BIOS can’t adapt, I’m going to try to score a motherboard with BIOS like mine from someplace.

If I crash & burn them, I’ll post from here and cry for help.

Off to see the wizzard…

[/que ‘evil monster gonna get you’ music]

14

Are you sure that the two hard drives have about the same transfer speeds? The RPM doesn’t tell the whole story. I remember that when I upgraded a hard drive in my windows machine a couple of years ago I was amazed at the difference it made in boot times.

15

Okay, selected DMA in as many places as I could in the Windows device manager. It was not on all cases on either computer.

Also went into BIOS and in 'Integraded Peripherals I can change all 4 IDE drives under PIO to auto, (where they are set) or mode 0 through 4, can not turn them off and just below that I can do all 4 IDE drives under UDMA in auto (where they are set,) or disable. Only tow choices.

Defragging drive on ‘IDE the second’ on SO’s computer and it is still slow…

??? Need to get a diff MB?? I fear trying to change the BIOS chip. I will ‘screw the pooch’ trying to do dat I’m sure.

More ideas or comments?

16

GusNSpot, does Windows say that the drive in question actually is running in “Ultra DMA Mode”? You can select “DMA if applicable”, but it’s what the read-only “Current Transfer Mode” field indicates that counts.

Any chance that LadyMack’s Win98 installation is spyware-ridden or has a virus? Do you have a clean “Ghost image” or suchlike for that drive that you could quickly reload to see if it’s hardware or OS-related?

[Of course, I know you’ve backed up all the data for both machines before you start playing around with Ghost or similar…]

Is her “slow HD” a recent thing, or has it always been that way?

I can’t imagine that trying to update the BIOS chip would help… DMA has been around for more years than the Athlon and Duron CPUs, so unless it’s a well-known bug (have you tried Googling the motherboard model numbers for previously-reported problems?) I’d say do it only as a last resort before tossing the motherboard.

Good luck and keep us posted…

17

Hers has always been this way. Ever since at least 2000.

I’ll keep checking and see if I can find an online PDF of the MB manual. I have one for my MB, but I have had no luck finding one for hers.

I’ll be back …

Gone to go look for this:

18

In Windows 2000, the only OS I have right here right now, it’s “Device Manager -> IDE ATA/ATAPI Controllers -> (Primary / Secondary) IDE Channel Properties” -> “Advanced Settings” tab. ISTR that it’s almost the same on Win98.

[Hmm… just felt an earthquake here in the Bay Area.]