Introduction

Recently in a forum I frequent, a comment was made about whether or not rounded cables offered the same performance of unaltered flat ribbon cables. Since I personally run all rounded cables inside of my system, I had my interest peaked, to say the least, as to whether or not this was actually true. So I decided I'd give it a test to see if it was true or not. I was hoping the data cited in the forum was inaccurate as they were quoting as much as a 60% performance drop between a flat ribbon cable and a hand rounded ATA-100 cable.

The Re-Write

After this article was first posted here, some very astute (and a couple less astute people who were too scared to register) made some comments on what I had written. Making some suggestions on what I should have added to the article and what I completely missed. One thing I will do for you guys is, I will NEVER write an article in notepad again. The lack of a spell check is quite annoying. Thanks to those of you out there who did catch my mistakes, I appreciate any and all useful and constructive feedback on stuff like this.

The Test

Well, before I go into my test here's some system specs to help you gauge what I'm talking about.

My Computer A.K.A. the test-bed:
Thunderbird 800 @ 800 (non-OC'd for testing purposes)
Asus A7V Motherboard with ATA-66 and ATA-100 controllers onboard
GF2 MX 32MB Video card
SB Live Platinum classic
6X DVD
CD-RW drive
256MB PC133 Micron SDRAM

Hard Drives:
IBM Deskstar 75GXP 30GB drive
Western Digital 30GB 7200 RPM ATA 100

System information:
Windows 2K Pro Original release version, no service packs or unnecessary updates installed.
Promise ATA-100 controller onboard running driver v1.0.0.0 off the Asus CD original

Now onto the tests:

The original test

The test bed was a non-partitioned Western Digital 30GB ATA-100 drive stationed alone on my Asus A7V's ATA-100 controller. HD Tach was used as the testing/benchmarking software. I ran both the read and write tests on it.

The cables used:

- Generic flat ATA-33 40 wire ribbon cable [no picture]
- Hand Rounded ATA-33 40 wire ribbon cable

(I had these handy for testing purposes, just to see what happened)

- WD OEM ATA-100 flat 80 wire ribbon cable
- WD OEM ATA-100 hand rounded 80 wire ribbon cable
- Machine Rounded ATA-100 80 wire ribbon cable purchased from FrozenCPU.com

Hand Rounded ATA-33 40 wire ribbon cable

WD OEM ATA-100 flat 80 wire ribbon cable

WD OEM ATA-100 hand rounded 80 wire ribbon cable

Machine Rounded ATA-100 80 wire ribbon cable purchased from FrozenCPU.com

For testing I ran HD Tach on the ata-33's once each, and twice each on the ATA-100 cables. Took screencaps of the results and saved them. Now onto the results.

The Results

Well, for the test I spent about an hour and a half one night swapping cables and all kinds of stuff like that. I think my motherboard no longer likes me now.

The Tests:

First I ran the Flat ATA-33 cable just for kicks to see how it'd perform in the test. No reason or anything, I just had one around and wanted to test it out.

Click To Enlarge

After than I rounded the flat ATA-33 cable just so nobody could say it was a different cable used for testing. This is the hand rounded one's results.

Click To Enlarge

As far as I know, nobody makes machine rounded ATA-33 cables, and even if someone makes them, I don't have any so they were left outta this test.

Then I swapped in the flat ATA-100 cable that came with a WD Hard drive. According to the other test run by somebody else, this should have been the fastest one of all. This was run twice.

Click To Enlarge

Click To Enlarge

After the flat one, I tested out a hand rounded cable I made a few months back from another WD Hard Drive, so while the cables are slightly different colors, I'm assuming they're close enough to the same cable.

Click To Enlarge

Click To Enlarge

Then for the final test I had purchased a machine rounded ATA100 cable just for this test from FrozenCPU.com. Its a nice cable and I liked it.

Click To Enlarge

Click To Enlarge

For those of you who prefer numbers in writing, here's the results of the tests:

Cable Type Random Access Time Burst Speed Minimum Read Maximum Read Average Read Minimum Write Maximum Write Average Write 33 Flat 13.6 30.1 18170 27410 22714 12949 27243 19836 33 Hand 13.6 29.9 18170 27270 22541 10929 24632 19568 100 Flat 1 13.7 75.9 18181 33262 24369 12962 26538 19730 100 Flat 2 13.6 75.8 18181 33256 24368 14124 26658 19745 100 Hand 1 13.7 71.7 18191 33256 24339 12957 26215 19870 100 Hand 2 13.6 72.4 18180 33259 24347 14335 26542 19751 100 Pro 1 13.7 76.0 18178 33287 24372 13699 26593 19689 100 Pro 2 13.6 76.2 18183 33263 24372 12971 26590 19823

The Second Gen Test: DVD & CD-RW drives stationed on my ATA-66 controller
IBM Deskstar 75GXP 30GB alone on Primary Master on ATA 100 controller
WD 30GB ATA-100 Secondary Master on ATA 100

Testing the IBM drive as it has a higher transfer speed, so it'll show more as far as cable quality possibly. Ended up re-testing the WD drive just for kicks when I was doing this.

OS: Win2K, with same setup as before
ATA-100: Onboard Promise ATA-100 controller with Driver v 1.0.0.0

Test Software: Once again HD Tach will be run, only this time I will NOT do write tests, as it can only be done on a non-partitioned drive, and I'm NOT about to reformat my system for this test. Deal with it ;)

IBM Deskstar test results:

Cable Type Random Access Time Burst Speed Minimum Read Maximum Read Average Read 33 Flat 1 13.7 30.3 11360.0 30177.0 26856.1 33 Flat 2 13.7 30.3 15014.0 30157.0 26723.1 33 Flat 3 13.2 30.3 6005.0 30145.0 26642.7 33 Hand 1 15.4 30.3 12102.0 30125.0 26856.3 33 Hand 2 12.9 30.3 9558.0 30129.0 26614.5 33 Hand 3 12.9 30.3 11303.0 30136.0 26797.5 100 Flat 1 12.5 77.3 11434.0 38663.0 28913.5 100 Flat 2 12.5 77.2 9159.0 38668.0 28832.2 100 Flat 3 12.6 77.4 9618.0 38658.0 28981.1 100 Hand 1 12.6 77.4 14604.0 38675.0 29079.7 100 Hand 2 12.4 77.2 12602.0 38658.0 28987.9 100 Hand 3 12.9 77.1 11004.0 38661.0 28987.7 100 Pro 1 14.8 77.5 13622.0 38670.0 28905.6 100 Pro 2 12.6 77.4 14606.0 38671.0 28995.5 100 Pro 3 15.4 77.6 11390.0 38667.0 29054.3

WD Results:

Cable Type Random Access Time Burst Speed Minimum Read Maximum Read Average Read 33 Flat 1 13.9 30.2 1817.0 27532.0 22752.8 33 Flat 2 14.0 30.2 18177.0 27563.0 22808.9 33 Hand 1 13.8 30.2 18179.0 27520.0 22797.0 33 Hand 2 13.8 30.2 18180.0 27508.0 22838.9 100 Flat 1 13.8 76.3 18179.0 33254.0 24370.5 100 Flat 2 13.9 76.4 18177.0 33263.0 24369.9 100 Hand 1 13.9 76.2 18181.0 33253.0 24371.1 100 Hand 2 13.9 76.4 18174.0 33266.0 24372.3 100 Pro 1 14.1 76.2 18186.0 33250.0 24371.4 100 Pro 2 13.8 76.3 18185.0 33262.0 24371.5

Conclusion

Well, here's what I came up with on the tests. There was a 1% difference roughly between the high and low-end performance on the ATA-100 cables. I hardly find this to be anywhere near the 60% drop that some other people was quoting. Surprisingly the machine-rounded cable from www.frozencpu.com bested even the flat cable slightly. Granted the difference would go under the "negligible" description, but it's worth noting. It probably has to do with how the cable was rounded.

What really surprised me was how well the ATA-33 cables did in the test, only coming in about 2MB/s behind the ATA-100. That was impressive. I'm almost tempted to think the whole 80 ribbon cables we're spending a LOT more money for are purely hype. But once higher throughput devices come out I expect the ATA-33 cables to remain at near where they were in this test while the ATA-100 picks up speed to match the new drives.

Now some people don't know why people round their cables in the computer. Well, the simple reason is 'airflow'. The flat IDE and SCSI cables are SO bad for airflow in a case simply because they're SO wide and flat. They get in the way of any kind of decent airflow. An added bonus to the rounded cables is that they're easier to move around inside the case than the normal flat ones are, meaning installation of them is simpler for you to do.

The conclusion, go ahead and round your cables if you've got the steady hand to do that. The ata-100's were difficult to do, but not impossible. The machine rounded ones available online are not cheap, but they obviously do a great job, and I say they're worth the money.

New Conclusion

Apparently Windows sucks as an operating system. Wait, that wasn't what this article was about…. Oh, wait; I know why I said that. The newer results you see in the test on the Deskstar are extremely lackluster in performance, and it's all due to the operating system. The major spikes you see in the Deskstar where the performance goes thru the floor, is when windows tried to do something while running the test. However the new test does help point out some things that are kinda useful. The ATA-33 cables don't slow much when using a lower performance ATA-100 drive even, as shown in the WD tests, they only lagged by about 2MB/s. Whereas, in the IBM tests, the high-end speeds were notably above what the ATA-33 cable could run. I'll have to remind myself that the next time I reformat my system, I can switch drives and do the test on the IBM where windows isn't raping it the whole time.

Hopefully this update helped clean up some of the issues there were with my last one. Special thanks to those who gave me some constructive feedback: rstarr the forum regular ;), diocco, bd134492, a couple of unregistered folks that gave good feedback, Honda (the dude from the [H]ardForum who kicked me into writing this thing), Nicholas Bodley and Chuck Tribolet for some serious tech feedback. Thanks for the feedback guys, if you feel like handing me any more, feel free!

-Mike

Re-Printed From SLCentral