Author Topic: Asus A7V266-E (Read 800 times)

mrsid2 · « **Reply #15 on:** March 31, 2002, 02:47:43 AM »

Minus: Judging from the reviews I've read of the 333 boards, they're not any faster than 266a and will have the same set of problems.

I use kt266a myself and it does have some issues with pci latency related operations. Fortunately my motherboard contained bios support for adjusting it so I no longer suffer from problems.

Still, the more I play with AMD based systems, the more I start to agree with Skuzzy. If you want cheap speed and hours spent getting to know why your system is not working properly - buy AMD. I don't mind this since this also relates to my work and goes from practise.

If you want stinky pricey bloated system which works like a charm when you slap the things on the board - buy Intel. You lose something in both ways.

Skuzzy · « **Reply #16 on:** March 31, 2002, 09:22:58 AM »

Actually mrsid2, FAT32 is not inherently any more of a security risk than NTFS, as long as the only things loaded in the network are the Ethernet drivers and the TCP/IP stack.
Yes, if you load anything else, such as NetBUIE, or MS Client, or have File and Printer sharing enabled, or install as part of a workgroup you do open yourself for some security problems.

I can hard crash a NTFS system in about 5 minutes, but then my environment is a pretty stressful one from a networking standpoint. My only hard crashes occurred using Exchange or IIS in stress testing. Sometimes the filesystem would recover, other times it required a full OS reinstallation. Reverting to FAT32, I would see lost files and such but always managed to get the system back up without an OS reinstall.
Both filesystems have plus and minuses. Just depends on your environment.

It's good to know XP gets around the file size limitation. Heck, for video work, I would go with a SCSI adapter and a Seagate Cheetah HD is you want really good video streaming performance. Expensive, but very fast.

mrsid2 · « **Reply #17 on:** March 31, 2002, 09:47:05 AM »

Skuzzy what about user restrictions INSIDE your network?
How do you manage confidential filefolders with fat32 in order to keep them out of reach of the general users?

Skuzzy · « **Reply #18 on:** March 31, 2002, 11:51:10 AM »

Inside the network is not an issue. First, we never enable file and printer sharing. Even is someone did, they would be all alone unless it was a conspiracy.
My standing policy is; if you alter your computer system without authorization (including the installation of any software), you will no longer be employed here. We run network monitors to see if that happens and do a periodical random check of the computers as well.
We do not have any workgroups enabled, save one, and it is on its own network subnet, and binds to tcp/ip for broadcasts which keeps the broadcasts from reaching any other part of the network. Putting that network behind a firewall makes it very hard to get to from the outside as well.

If you cannot see the drive, you cannot get to the files irregardless of permissions. We download and test various hack programs to make sure all is secure as well.

I periodically have to notify clients to turn off NetBUIE on thier networks so they can remain invisible. Most people seem unaware you can bind a client to TCP/IP on a subnet and the broadcasts do not cross outside the subnet.

That answer your question?

mrsid2 · « **Reply #19 on:** March 31, 2002, 04:04:43 PM »

Well, kinda, thanks..

I guess you have no need for a shared network.

bloom25 · « **Reply #20 on:** March 31, 2002, 09:01:04 PM »

MrSid, I'm sure I've mentioned this before, but "pci latency" issues are not an "AMD" specific problem. Intel i850 and i845 boards also suffer from this problem. Your problem is actually a VIA issue, not an AMD issue. (P4X266 suffers from it as well.)

I think there is a lot of confusion on this issue:

The problem is due to the 133 MB/sec bandwidth limitations of the PCI bus itself and how it behaves when that limit is exceeded. Intel's workaround was simply to limit bandwidth usage to 80 or 90 MB/s. (I can't remember the exact figure at the moment.) VIA's fix was to change the interupt timings to reduce the bandwidth used at any given point. Both solutions (of course) reduce performance. I850 and i845 hard disk benchmarks are, as a result, pretty poor compared to other chipsets.

Some AMD supporting chipsets, nVidia's nForce for example, has no reported "pci latency" issues. I don't know if the Ali and Sis chipsets suffer from this.

The SB Live series of soundcards had driver issues that could cause some serious problems. Even though 133 MB/s sounds like plenty, remember that on all but a very few examples, IDE disk traffic was transfered across the PCI bus. Burst transfer speeds for some drives may require all available bandwidth. (ATA133 can support a transfer rate of 133 MB/s remember.) Add a SB Live playing an audio stream while transfering large files back and forth between the primary and secondary IDE channels and disaster resulted. I'd imagine a raid array would be even worse. (Nvidia got around the problem by moving IDE traffic to a "Hyper Transport" 800 MB/s capable bus.) I personally think that Serial ATA should help with this problem when it becomes available later this year. (If it's not delayed again... )

In no way does the processor itself directly cause this problem.

_____________________________ ______________________

I personally generally go with NTFS on my own system for security reasons, but I will say that in a dual boot configuration with Win9x, FAT32 is more convienient as Win9x cannot read NTFS formatted disks.

If you turn off some of the silly Windows 2000/XP disk indexing services, NTFS also improves disk performance compared to FAT32 on large partitions.

I'm not sure if the 4 GB limit is a FAT32 issue or an OS issue. I personally think it's probably an OS issue.

In Eagler's case, I'd probably suggest FAT32 IF he planned on a dual boot configuration where Win9x was heavily used. You *could* set aside a small special partition for Win2k/XP OS itself and format it NTFS. I think I'd probably go with FAT32 though. I've actually had a power failure wipe out an NTFS formatted disk to the point where I had to reformat it.

mrsid2 · « **Reply #21 on:** April 01, 2002, 01:13:17 AM »

Bloom25: I know about the pci latency issues with Intel boards also. What I meant really was that I haven't seen an amd based system yet that would have worked 'out of the box.'

They need more or less tweaking to have optimum performance, or sometimes _any_ kind of performance, something I usually can skip if I use an Intel board.

I bought my first amd with abit amd760+686b board and tried to get it stable for 3 months, losing a hefty amount of valuable data in the process even though it was due to a broken harddrive. The 686B systems had the nice 'feature' of corrupting large data transfers so it was a pleasure making a backup from the 30 gigabytes of data I had stored. Read impossible. I also suspect the 761 board was faulty even though the reseller never admitted it.

Anyway, after 3 months I changed the board to a 266a one and even this one hasn't worked as expected from the reviews. I've had to toy with the settings to get IO to work fluidly with multitasking.

Not even you can convince me that this system would have been even a fraction as stable like my old BH6 was.

Now with my IBM 60GXP giving me noises similar to the 'click of death' I'm really starting to get frustrated with this box.

Judging from my experience I can reckommend AMD to people who want high performance and low cost - if they're willing to tweak the system. If they're n00bs.. The choice does get harder.

Lephturn · « **Reply #22 on:** April 01, 2002, 08:42:22 AM »

Quote

Originally posted by bloom25
I personally generally go with NTFS on my own system for security reasons, but I will say that in a dual boot configuration with Win9x, FAT32 is more convienient as Win9x cannot read NTFS formatted disks.

If you turn off some of the silly Windows 2000/XP disk indexing services, NTFS also improves disk performance compared to FAT32 on large partitions.

I'm not sure if the 4 GB limit is a FAT32 issue or an OS issue. I personally think it's probably an OS issue.

I can tell you absolutely for certain that the 4GB file size limitation is a FAT32 file system limitation. If you look at how the file system is organized, the math says you can't have any larger than 4GB in a single file.

http://www.microsoft.com/windows2000/en/server/help/default.asp?url=/windows2000/en/server/help/choosing_between_NTFS_FAT_and_FAT32.htm

NTFS can have a partition up to 16 Exabytes, and theoretically you can have a single file as big as the entire file system. Practically, I'd say a single file of a Terrabyte or 2 would be the most I'd want to try.

I work doing performance testing on NT/Win2k for my company. I have not had a single problem with NTFS. I don't know what the heck Skuzzy is doing with it, but I believe for your average user NTFS is far more stable and secure. I can't remember the last time I actually had to reformat a drive in NTFS due to errors... it's just never happened to me. I generally migrate 50 GB file structures and up to 100,000 users using our products, and I've never had NTFS fail. Ever. Obviously I don't do the kind of file I/O that Skuzzy tests with, but then again I doubt any of you guys would either. FAT 32 is a hack to make FAT better, but it's not nearly as stable, efficient, scalable, or secure as NTFS in my experience. NTFS's file level security just makes it a must-have in my book, but I think I'd use it anyway.

BTW, this is a great article comparing the two file systems: http://www.anandtech.com/guides/viewfaq.html?i=63

bloom25 · « **Reply #23 on:** April 01, 2002, 05:08:50 PM »

I should have just done the math. Fat 32 uses 32 bit addressing, meaning 4294967296 Bytes is the limit. Since Windows considers 1024 KB as a MB, that would come out to around 4.1 Gigabytes.

I guess if you do need to have files larger than 4 Gigs, than NTFS is the only way to go.