Accelerate Your Mac!
Daily News Updated Several Times a Day!

= Current Site Feature =
SCSI Report Part 3: SCSI Raid Comparisons
(Single Card vs. Dual Card w/5 Software Packages!)

= Complete Review List =

-Send your News items, Tips and Guest Reviews. -
Advertise here! - Help support Accelerate Your Mac!

Last Update: 4:02PM EST
02/25/98 - Wednesday's News:
  • Adaptec's latest readme notes: Here's a few items of interest from Adaptec's latest readme file:

    " The PowerLogix PowerForce 'G3' processor upgrade cards are not currently compatible with various PCI cards including your PowerDomain card. The problem occurs before the computer gets to our card so we are unable to work-around the problem.

    If you have a computer with a Tanzania-based logic board (such as the Power Macintosh 4400), you may need a patch from Apple for your board to work (the fix may be preinstalled or part of future systems). This is an extension which updates 'OpenFirmware', so once youšve started with that patch active even once, you can startup with extensions off. The patch may be named 'OpenFWBridgeFix'. If you reset NVRAM (typically as part of trouble-shooting) youšll need to startup with extensions on to reapply the patch.

    If you have a PowerComputing PowerCenter Pro series (now supported by Apple Computer) you may need a patch called 'PCI Timing Update' for your board to work. It is an extension which makes changes to 'OpenFirmware', so once youšve started with that patch active even once, you can startup with extensions off. If you reset NVRAM (typically as part of trouble-shooting) youšll need to startup with extensions on and any boards affected by this problem removed to reapply the patch.

    In the Umax S900, you need to put your SCSI board in one of the top two slots.

    If you have a video capture board, or have trouble with pauses or stuttering (audio, writing data, et cetera) when moving information to and from devices attached to the PowerDomain card, please read the section on 'Using the Video Buffer Settings'.

    If you have a Radius PCI VideoVision, please make sure you have their latest update. Check http://www.radius.com/ or call their Technical Support department at 408-541-5700. "


    Adaptec 2940UW Bios v3.0 Feedback: Henry Goh of Singapore writes in that bios version 3.0 caused immediate problems for him. Since the PowerForce/Adaptec 2940 (or 3940) combination does not work here with wide drives at all with either bios version (the drive will not mount) I'm curious to know what his configuration is (narrow SCSI drives I assume). Henry wrote:

    " I wish to bring to everybody's attention that the Adaptec BIOS flash V3.0 does not work with my PowerForce G3 250/250 card.

    I downloaded and flashed the 2940UW card and then restarted the system. The system complete hung during boot up. Tried everything from zapping PRAM etc with no avail. Even tried booting from CD-ROM...dead end.

    Had to reinstall my old 604e/180 MHz CPU to restart. Then I flashed the BIOS back to V2.5.

    Had I sold off my 604e crad, I would have been in deep trouble.

    I thought everyone should know about this as it is impossible to restart if you do not have a 604e card to reflash the BIOS. "

    I'm well aware of the PF/2940 issues, so please no mail about it. I posted this in case there are any other readers that are running the combination fine (rare if you have wide scsi drives in my experience) as a warning that updating to the v3.0 bios may cause problems. I can attest to the fact the v3.0 Bios update was not written to address the PowerForce G3 issue. In the past a few users have reported the combination (2940v2.5/PF) does work for them, but not for me and most other owners of the two cards.

    Although the ATTO, Initio Miles and Streamlogix Jackhammer cards seem to work fine here with wide drives and the PowerForce G3, my 2940UW or 3940UW will not mount a wide drive when running in a Mac with the PowerForce G3 card installed.

    I've reported on this issue many times in the past - and v3.0 will not help those PowerForce G3 owners with problems. Version 3.0 contains interrupt code changes to work better with OS 7.6 and later Adaptec told me. Performance of v3.0 is a mixed bag compared to v2.5 as shown on my latest SCSI Raid tests


  • Virtual PC 2.0 Ship Date: Peter Koplan sent the latest info on VPC 2.0's release:

    Virtual PC 2.0 is scheduled to first start shipping on Monday, March 9. There will be three flavors of the product: Virtual PC 2.0 with Windows 95 (Estimated Street Price $149), Virtual PC 2.0 with DOS (ESP $49) and Virtual PC 2.0 Upgrade from 1.0 (ESP $35). The upgrade will be shipped as a flatpack, not a boxed unit.

    Can't wait - I'll be able to compare the 3dfx speed/support of RealPC and VPC with all the 3Dfx PC games here. It will be a welcome break from the SCSI data recording routine!


  • SCSI Comparison Update: I was up all night writing the final set of SCSI Raid Comparison Tests pages. This third installment covers single SCSI Card Raid tests with all 5 of the most popular Raid software packages (Anubis Raid, ATTO Raid, Conley SoftRaid 2, Remus Lite 1.4, and Raid ToolKit) and each of the 4 Ultra-Wide SCSI cards (ATTO Express PCI, Adaptec 2940UW, Initio Miles and StreamLogic JackHammer). I've included the dual SCSI card results in the tables to allow easy comparisons. Although this made the tables longer, I felt it was worth it. I've also added tab links to the top of each page (on all three reports) for easier navigation. I've not color coded the highest/lowest scores in the tables yet but I'll do that later today.

    As you'll see from the results of most tests, it seems that a single SCSI card delivers about the same performance as a dual SCSI card setup (at least with a two-drive array and 180mhz CPU). With more drives in the array and/or a faster CPU the dual card setup may show more of a performance advantage. With 4 or more drives in the array I'd expect the dual card setup to be a marked advantage. If I can obtain 2 more Atlas II 2.2GB drives I'll test for this at a later date.

    Although the ATTO benchmark showed a substantial increase in peak R/W speeds with the dual cards, the sustained R/W rates and most other benchmarks showed literally identical performance. Again, more drives would likely show more of a delta in favor of the dual card setup I suspect.

    The three sets of test results (single card/single drive, dual SCSI card/dual drive Raid, and single card/dual drive Raid) have a lot of valuable information - I know of no other review of Mac SCSI hardware and software that is as comprehensive anywhere on the Net or in print. I've not written the summary page yet and I'd like to repeat the Time Drive tests in some cases, but I'm going to take a break from SCSI for a few weeks to work on other reviews that are backlogged. The Apple G3 system pages, MacPicasso 540/3D0 and XLR8 G3 CPU card reviews have been delayed for too long by this extensive series of SCSI tests.

    I'm hoping to get some assistance in graphing the results in the future. This would allow an easier 'at a glance' indication of performance between all the variations. If there is anyone out there would like to do this for publication here, please contact me.

    Thanks again to MacGurus for all the loaner cards and RTK software. I also want to thank ATTO (ATTO Raid), Adaptec (2940/Remus Lite), Initio (two Miles cards), CharisMac (Anubis Raid) and Conley (SoftRaid 2) for submitting products. Without their support this project would not have been possible.


  • VST ZIP drive update: VST is sending a replacement ZIP drive for my PowerBook G3. They called me to say that my problems with drive clicking, etc. were not normal and they wanted to replace the drive. I think Iomega could learn a LOT from VST in the customer service department.

    The VST ZIP drive saves me space and time by allowing large file sharing via between the many machines here and 100MB removable storage on the road without packing a bulky external zip drive, cables and power supply. For systems under test I don't run a network, as it affects performance and the ZIP is my 'SneakerNet' file transfer protocol.


  • New Mac 3D Card: Philippe Braun sent news of a new Mac 2D/3D card based on the 3DLabs Permedia 2 graphics chip.

    He writes:

    " thanks for your always-informative (and sometimes really entertaining) web site. Since european products seem to have something of a hard time getting into the US, I thought you might not know and be interested in the new G-Rex 3-D accelerator from Phase5 in Germany.

    It uses the Permedia2 chip set, and costs DM 499 in the 8MB SGRAM version - a lot cheaper than any competing (Mac) product, and actually almost the same as the Wintel (Diamond FireGL1000 PRo etc) cards.

    Here's the URL:

    http://www.phase5.de/news/grex3de.html "

    I had rather dismal performance during testing of the original Permedia chip based Radius Tempest, but hoped that was all driver related. The Permedia 2 is said to be faster at both 2D and 3D however.


  • Voodoo II card update: Barring delays, I'll have a 12MB Voodoo II chipset based card by the weekend. After verifying it works in the Pentium II 300, I want to test it in the Mac to see if it will run with the original Mac Glide (3Dfx game) drivers [It doesn't].

    While on the subject - Jeremy Burden sent in the URL of a site that sells used 3Dfx cards of all brands. Evidently the new Voodoo II has prompted a mass sale of previous cards. The URL is: http://www.bushido-dojo.com/vdx/

    Always exercise caution when buying from individuals on the Net - and be aware that Voodoo Rush chipset cards will not work in a Mac (but the Monster 3D, Orchid Righteous 3D, and Pure3D cards do). The Voodoo Rush was 3DFx's later version of the Voodoo chipset that does 3D in a window, but comes as part of a 2D video card which won't work in the Mac. Typically these cards use the Alliance ProMotion 2D chip.

    The cream of the crop in original Voodoo chipset cards has been the 6MB Pure3D. Although the Canopus Pure 3D FAQ says they don't work in a Mac (there were no Mac 3Dfx drivers originally) owners have reported it does work with the full Quake drivers in my Video card forum in the past.


  • Umax S900 Motherboard Info: Besides the unique PCI bridge chip and dual CPU card slots (which go to waste typically, as their proprietary 2nd CPU card has been expensive), the Umax S900 also has 16MB of ram soldered on the motherboard. Ryan Dumperth wrote in to say he found out this 16MB cannot be cached - which surprised me. The Umax S900 motherboard is the basis for many 6 PCI slot clones on the market (PowerTools, MacTell, etc.). The full story of his discovery is detailed in his mail to me:

    " I grabbed version 1.5 of DayStar's nPOWER Control app (found in this disk image

    If you haven't seen it before, it has a way cool interface, and gives some salient info on your hardware setup. [As noted on my MP Problems Page, I found that tool last summer - it has a great memory test that picked up errors that Newer Tech's ram gauge missed-Mike]

    One awesome addition is the ability to turn off the L2 without rebooting. It works flawlessly on my S900's 512K built-in L2. TechTool Pro is the only utility which was still able to access my L2. It was totally disabled for all normal apps. [I've not tried it personally, but some readers have tried to disable their L2 cache with it for use with their PowerForce G3 cards-Mike]

    MacBench 4 showed this:

    1. S900/200; OS 8.1; 96MB of RAM; LibMotoSh; 7200 rpm drive on 10MB/sec internal SCSI, 4096K disk cache.

    2. Same, with L2 disabled.

    • Processor: 479/326 (47% increase with L2 enabled)
    • FPU: 655/590 (11% increase)
    • Disk: 398/307 (30% increase)

    This probably mirrors what would be seen in a similar 9500, but since I can't upgrade/remove my L2, it was very interesting to see what effect it has. One very disconcerting revelation from the program is that, apparently, the 16MB of RAM that is built onto the motherboard is not cached by the L2. I have 80MB in my DIMM slots, and the L2 part of the app shows that only 80MB is cachable. It's even worse because, according to MemMapper, my entire System Heap is sitting in the lowest 16, and I'd imagine that's a pretty busy part of memory. [That's what occurred to me as well-Mike]

    I'd be very interested in seeing an identical 9500's score, since they're pretty close, other than the built-in memory (which nPOWER Control does confirm is interleaved).

    I thought the L2 disable-feature might be of some use to you in your benchmarking. I don't know if any other programs can do that or not. Perhaps it could be another option for getting more accurate benchmarks of disk subsystems. Thanks for the great resource. I check in daily. "

    Thanks for the info Ryan. Of the Tsunami based (6 PCI slot) clones I prefer the PowerTower Pro - as it has a removable 1MB L2 cache, and although it has only 8 Dimm slots, that's enough for a total of 512MB with 64MB dimms.


  • More news later in the day.


Yesterday's News

Please keep this site alive by supporting my Sponsors

= Top of Page =

To reset frames - click here

Web Hosting sponsored by:
Dax Media

and
CornerStone Networks.


Copyright © , 1997, 1998.

No part of this site's content is to be reproduced in any form without written permission. All brand or product names mentioned here are properties of their respective companies.

Legal: Users of this website are bound by the Terms and Conditions of Use.