China DOS Union

smartdrv 10240 10240 means creating a 10MB high-speed buffer, right? But why is the speed when using Ghost on the hard disk basically the same as before? (I already set the hard disk block transfer mode in the BIOS.) What should I do to speed it up?

Not answering this for now. Give me back the 1 million first, then I'll tell you.

I didn't expect you to want my money too, just like PlayBoy. Is being rich a kind of mistake too?

Answering Chairman Mao's call: “Strike the local tyrants and divide the land” — down with the entire bourgeoisie!!!!!!!!!!!!!!!

I didn't expect possessing a jade disc to be a crime? I've decided to find a time to hold a “money-scattering convention,” and then write at the entrance “Chairman Mao and Master Rushi are not allowed in,” haha~~ because both of you hate money!!

Who said that? I love money the most.....

GHOST seems not to use SMARTDRV. I've tested it myself; it even seemed a bit faster without SMARTDRV.

The following is quoted from 冰水遥 on 2003-4-29 21:14:59:
smartdrv 10240 10240 means creating a 10MB high-speed buffer, right? But why is the speed when using ghost on the hard disk basically the same as before? (I already set the hard disk block transfer mode in the BIOS.) What should I do to speed it up?

A buffer is not the bigger the better. A buffer only helps speed things up when you repeatedly access files you've accessed before. On the contrary, if the buffer is too large, not only will it fail to make the system faster, it will actually reduce speed. Because every time before you access a new file, the computer first searches the buffer. You can't possibly always be running programs you've run before, right? That leads to a very low buffer hit rate, while also wasting a lot of time searching the buffer. As a result, system performance drops. I used to read many articles introducing disk buffers; in those tests, the best size was 2M (2048). If it's 2 or 3M larger than that, system performance won't improve, and with 10M like yours, system performance will definitely drop. Since current computers are much faster than they were back then, the performance drop is not obvious. If you want to test it, you should use older P55C or P54C (MMX Pentium/Pentium) machines. Then the performance difference will show.
When GHOST creates or restores a system, it never repeatedly reads the same files from beginning to end. It keeps reading new data, so high-speed caching is almost meaningless in that environment.
Disk transfer rate mainly depends on your motherboard design. It's basically an inborn issue, and it's hard to make major improvements afterward. Take the 845PE for example. I've used two models: the Gigabyte 667 Titan and the MSI 845PE MAX. With all other software and hardware the same, the former could reach a disk transfer rate of 320-350M/MIN in GHOST with compression enabled, while the latter could only barely reach 150M/MIN even during decompression. With a difference that big, how could a buffer possibly make up for it?

Hehe............ poster on reply #8, you just told him like that? He still hasn't brought out the money yet...............

I don't think what you're saying is necessarily right!!! Although I still can't say why! I once used three identical machines to do GHOST. Let's call these three machines A, B, and C. (Exactly the same Haier brand machines: Celeron 500 CPU (Mendocino core), 64M RAM, Quantum 20G lct20 hard disk (because of the hard disk cable, it only supports up to DMA33), motherboard chipset Sis630/540, south bridge Sis 5595.) When I cloned from A to B, the speed could reach 250M, but from B to C it could only reach 100M. That's far too much of a difference. Before that, I had already made the hard-disk-related settings in CMOS exactly the same too. A speed of 100M I could already reach on a CYRIX200, 32M RAM, Samsung 1.08G, VIA　VT580!! So where exactly is the difference? Also, on the same Haier brand machine, I previously made a C drive image. When I booted with a normal boot disk and did a partition restore, it could only reach at most 50M, but when I booted with a boot disk containing SMARTDRV, it could reach 100M. This is an indisputable fact.

冰水遥
Your testing method is wrong. For example: “but when I booted with a boot disk containing SMARTDRV, it could reach 100M. This is an indisputable fact.”
What you should do is remove the SMARTDRV from that disk and try once more. I believe it will definitely exceed 100M.
nre's reply is correct, too bad you don't take it seriously.
Judging from your reply above, your test has no meaning at all.

I am taking it seriously. Also, when you say to remove that Smartdrv.exe, isn't that just the first case I mentioned (booting with a normal boot disk)? I'm wondering whether the buffer size was set wrong, otherwise how could the speed on the same computer differ by a whole factor of two just from using different boot disks?

Is no one going to continue discussing this? How exactly can smartdrv speed up GHOST (I think it also has something to do with the hard disk's cache; the Quantum ＬＣＴ　２０ 20G hard disk actually has only 128KB cache). (It definitely has something to do with the motherboard. On this ASUS mel-m 440lx chipset motherboard — supports ATA33 — with 128M RAM, and two hard disks, namely a Samsung SV2001H 20G------single-platter 40G, 2M cache, 5400 rpm, supports ATA100, and a Quantum ＬＣＴ　２０ 20G----single-platter 20G, 128K cache, 4500 rpm, supports ATA100, the speed can reach 450M/min.)