Mau1wurf1977 wrote:

What did you run? SiSoft Sandra?

If it can also test a secondary drive I test all sort of drives I got floating around 😀

Yeah, Sandra 99 I believe.

I'd use ATTO Disk Benchmark and look for at small write performance because that is what causes problems. Might need to run it on a modern OS. If you see <1MB/s then not good.

swaaye wrote:

I'd use ATTO Disk Benchmark and look for at small write performance because that is what causes problems. Might need to run it on a modern OS. If you see <1MB/s then not good.

Yeah, I believe ATTO can only test writes on an empty drive right?

Artex wrote:

Yeah, I believe ATTO can only test writes on an empty drive right?

No, it will run write tests on a drive with data on it. It won't harm the data either.

swaaye wrote:

Artex wrote:

Yeah, I believe ATTO can only test writes on an empty drive right?

No, it will run write tests on a drive with data on it. It won't harm the data either.

Hmmpf...the versions I have (2.60, 2.61) complained that the drive had to be empty, although these are very old versions running in Windows 98SE.

Mau1wurf1977 wrote:

Too much talking, I want to see some benchmarks of your storage solutions 🤣 ...

TELVM wrote:

http://s9.postimg.org/t75bqv74f/Crystal_D_HDD_ATA33_NTFS_Tualatin_1470_140_MH.gif […]
Show full quote

Mau1wurf1977 wrote:

Too much talking, I want to see some benchmarks of your storage solutions 🤣 ...

Ok, gotta ask, but how did you get the Tualatin 1400-S to work on a 440BX board??? I ask because I have a 1400 that's been sitting idle for yeeeears. Got it for $5 from a local computer recyclers processor bin and never got a board for it (tried it in my i820 and some random Via based socket 370 board I had).

So.. basically SSD FTW. Look at those 4K CF write numbers... argh. I ordered two 30GB OCZ Vertex Turbos - hopefully that will be the end of my problems, even if it's the slowest SATA 1.5 version.

misterjones wrote:

Ok, gotta ask, but how did you get the Tualatin 1400-S to work on a 440BX board??? I ask because I have a 1400 that's been sitting idle for yeeeears. Got it for $5 from a local computer recyclers processor bin and never got a board for it (tried it in my i820 and some random Via based socket 370 board I had).

A) Slot 1 Adapter?
B) Pin-modded Processor?

Artex wrote:

So.. basically SSD FTW. Look at those 4K CF write numbers... argh ...

SSDs just cream everything else (short of exotics like self powered ramdrives) but don't be deceived, @ UDMA-2/ATA-33 a CF feels quite faster than a spinner, courtesy of the eight times higher 4K read speed and negligible access times.

misterjones wrote:

... how did you get the Tualatin 1400-S to work on a 440BX board???

Long story short you need:

· A late production 440BX board with VRM 8.4 compliant power controller (earlier 440BX boards with VRM 8.2 for Pentium II won't work).
· A slotket with voltage clamping chip (like the MSI-6905 v2.1/2.3 MASTER).
· Some tinkering on the Tualatin CPU pins and/or socket 370 at the slotket.

I have an Upgradeware Slot T which lets you use Tualatin on any 440BX with the correct VRM. In my case, Abit BF6.

Guys, I'm not using a BX440 board 😀

For Windows 98SE Crystal Mark can't be used. Because this Thread is about W98SE and disk performance, we should use tests that run under W98SE, right, makes sense? 😀

Link to ATTO please. To the version that can do write. Because if we can't agree on some benchmarks, this thread isn't gong anywhere.

Ok SiSoft Sandra is a bit of a pain as the details of the benchmark result are in the list at the bottom and you can't display it all on one screen 😀

So I left out the last information in the list, which is access time, which was 0 anyway.

System specs:

Here is a 32GB Kingston CF card:

Same CF in ATTO:

And here a 4GB Seagate Microdrive.

Remember I test drives on the secondary channel, so that the tests don't interfere with the OS which is on the primary channel.

Very consistent and seems to be limited by the drive mechanics.

And here we have a 320 Western Digital Scorpio Blue SATA notebook drive. PCI SATA Controller is that typical Silicon Image I use since day 1 (cheap on eBay). Driver for it was from the SI website.

As I always do, it has a single 30GB (30720 MB) partition it it.

And finally a Sandisk Extreme 120GB

Findings

Well I believe this ATTO benchmark is definitely a keeper. It confirms what you guys have been saying, CF cards struggle with writing small files. Also reading small files is quite poor.

Quite impressed with the Microdrives, however this benchmark doesn't show access time and liner speeds. So I think we will need to add another benchmark to show this!

Notebook HDD seems to also do quite well. But again, access time doesn't show up in ATTO.

Once again Phil, thanks for all of your hard work! I'm going to stick with my Notebook drives & SSDs.

Artex wrote:

Once again Phil, thanks for all of your hard work! I'm going to stick with my Notebook drives & SSDs.

No worries 😀

I'm changing my W98 bench back to the BX-2000 and using a notebook HDD now 😀 The Microdrive felt a bit slow 🤣

The Microdrive results are interesting. It looks like it could be bottlenecked by an interface. HDDs don't usually have such flat performance.