Widescreen Gaming Forum

On your third point, Cyna... that NAS was advertised as a gigabit ethernet connection, and it's ethernet only... it reports 1000Mbps in the speeds and various setup pages, so I have no reason to believe it isn't. If the random access times are that poor... ouch. Oh well. To be honest, once it's full of stuff, I'll only be reading files off it occasionally. Obviously I'm just destined to get annoyed at how long it takes to write to. ;)

Hit up newegg, check out all the TB and 500gb drives there, a 500gb's average read/write latencies are 8-9ms/10-13ms respectively. That's just the base response times, now if a file is fragmented, that increases the time as it has to seek across the drive to get the data. A TB drive's average response times are ... wow, they don't even want to list them, they must be bad then... the best I saw was a seek time of 8.9ms, which is how long on average it takes the read arm to position the heads to read the data you want, now if the data was in a long string, that's one seek, and just adjusting with the line to read the spiral of data. Fragmenting makes that 9ms here, read some, 9ms, read some, 9ms, read some, you see how that adds up, especially on a REALLY fragmented drive. Not sure how that NAS handles fragmentation, but if it just lets the drive do it's thing to reduce the firmware OS size, then it's going to be severely fragmented to hell...

_________________
Current build: Intel i5-4670k | 32GB DDR3 | RAID0 2x250GB 850EVOs | nVidia GTX980Ti (MSI G1 Gaming Edition) | SoundBlaster Zx | Dell 2405FPW (Landscape, primary) & HP w2338h (Portrait)

WD's page lists that nas as 10/100/1000. Though listing it as gigabit and performing as such is quite a different things, I've got 5+ gigabit nics, all from different chipsets/manufacturers, so far, few run more then 200mbits/s. There seems to be a general wall whenever we leave 10/100 standards and go into the gigabit hack-standard.

Cynagen, my practical data counters your numbers, I've done 55mbytes/s avg from a single IDE / ATA133 drive ...

And unless you've got a near full drive with countless times of erasing/rewriting large amounts to it, you can forget about fragmentation.

Fragmentation happens only when no space allows for a file to be written in the same spot. On near/half empty drives, it's about pointless.

Case in point, this drive I've been using for years has about a dozen files with even a few fragments, mostly small windows temporary files.

WD's page lists that nas as 10/100/1000. Though listing it as gigabit and performing as such is quite a different things, I've got 5+ gigabit nics, all from different chipsets/manufacturers, so far, few run more then 200mbits/s. There seems to be a general wall whenever we leave 10/100 standards and go into the gigabit hack-standard.

Cynagen, my practical data counters your numbers, I've done 55mbytes/s avg from a single IDE / ATA133 drive ...

And unless you've got a near full drive with countless times of erasing/rewriting large amounts to it, you can forget about fragmentation.

Fragmentation happens only when no space allows for a file to be written in the same spot. On near/half empty drives, it's about pointless.

Case in point, this drive I've been using for years has about a dozen files with even a few fragments, mostly small windows temporary files.

Was that 55mb/sec sustained for more than a second or two? And i'm not talking about the window's screen updates, i'm talking actual seconds. Vista likes to post an average, and the first ones are usually high because of the burst of data written/read right off the bat, then it calms down to the average.

_________________
Current build: Intel i5-4670k | 32GB DDR3 | RAID0 2x250GB 850EVOs | nVidia GTX980Ti (MSI G1 Gaming Edition) | SoundBlaster Zx | Dell 2405FPW (Landscape, primary) & HP w2338h (Portrait)

Well, I'll do a bit more experimenting when I've got some time... maybe try setting up a RAM disk in a testing OS to remove the HDD speeds from the equation.

Thanks for continued input. :)

Paradigm Shifter:I'm still looking for solutions myself, maybe going back and forth we'll both get our gear working 'as advertised' good luck.

Cynagen: Burst part was over 70, sustained transfer of a 2.3gbytes file at 50-60mbytes/s and since windows typically has the lower number way of displaying data sizes, it's likely a bit over that.

I wonder whether the increased speed in Vista is due to it's better caching system? Or rather, it's propensity for chewing up RAM by caching as much as possible. But for transferring large amounts of data, it would make SFA difference.

I thought hard drive speeds were in MegaBytes/s.

(Source)
As noted above: these are the theoretical maximums of the interface (not the drives).

This is echoed at Wikipedia:

A typical desktop HDD, might store between 120 and 300 GB of data (based on US market data), rotate at 7,200 revolutions per minute (RPM) and have a media transfer rate of 1 Gbit/s or higher. (1 GB = 109 B; 1 Gbit/s = 109 bit/s)
...
The fastest “enterpriseâ€

My theory why vista2vista works as it should is because they (finally) redid the (censored) networking systems that have been unstable since at least w2000. It also has more auto-explore/auto-configure settings that are otherwise fixed/set in 2000/xp like the tcp receive window, mtu, etc.

Could be their driver signature required just kicked network maker's rear into making better drivers too, who knows.

Still, it should function the same regardless the os involved, especially under the same hardware.

And Tyinsar, yeah you got it right. Just like usb's theoretical max of 480mbit is 60mbytes.

I thought hard drive speeds were in MegaBytes/s.

Code:

              Transfer       Transfer
   IDE              Rate           Rate     Pins
   Drive     PIO   MBytes   DMA   MBytes     in
   Type      Mode  per sec  Mode  per sec   Cable
   ATA       0      3.3     0      4.2      40
   ATA       1      5.2                     40
   ATA       2      8.3
   ATA-2, 3  3     11.1     1     13.3      40
   ATA-2, 3  4     16.6     2     16.6      40
   ATA-4 (ATA-33)           2     33.3      40
   ATA-5                    0     16.6      40
   ATA-5                    1     25.0      40
   ATA-5 (ATA-33)           2     33.3      40
   ATA-5                    3     44.4      80
   ATA-5 (ATA-66)           4     66.6      80
   ATA-6 (ATA-100)          5    100.0      80
   ATA-7 (ATA-133)          5    133.0      80
   Serial ATA (SATA)        5    150.0       4
   Serial ATA II (SATA II)  5    300.0       4

This would mean that SATA2 would be communicating @ 2.4gigabit/s... SATA @ 1.2gbps, ATA-133 @ 1.064gbps, ATA-100 @ 800mbps, etc...

Those numbers just, don't seem realistic. Why would they spend the time and money making these ports as fast as this when there's never going to be a harddrive (short of DDRRAM/SSD drives) that will ever be able to saturate those lines.

Edit: My 400th post, yey?

_________________
Current build: Intel i5-4670k | 32GB DDR3 | RAID0 2x250GB 850EVOs | nVidia GTX980Ti (MSI G1 Gaming Edition) | SoundBlaster Zx | Dell 2405FPW (Landscape, primary) & HP w2338h (Portrait)

This would mean that SATA2 would be communicating @ 2.4gigabit/s... SATA @ 1.2gbps, ATA-133 @ 1.064gbps, ATA-100 @ 800mbps, etc...

As you noted earlier those are just the theoretical max speeds of the interface. Even at that it hasn't been that long since consumer level drives could saturate even the ATA-66 standard (at least from what I've read).

To me it's kind of like graphics cards switching to PCI-e: at the time graphics cards really only used about half of the maximum bandwidth AGP gave them - we didn't need it yet when the switch came.

To me the switches (AGP->PCIe & ATA->SATA) seem to be partly about future potential and partly about marketing (specifically getting us to buy new stuff).