Sign in with
Sign up | Sign in
Your question

Highest MB/s transfer rate on a home LAN?

Last response: in Networking
Share
May 30, 2006 7:21:32 AM

Edit: For you new guys to this topic, please post your highest MB/s transfer rate between 2 PCs on ur home LAN or Network. Thanks

Hey guys. Please HELP! So I wanted to upgrade my home network to transfer files between 4 PCs all in differnet rooms. My family shares files all the time like videos, mp3s, huge WAV files for editing, games, photos. and for example i gave my brother my whole music folder which was 10GB. Well it took over 30 mins or so to transfer everything, while rates varied between 5-10MB/s. My PC is the base sation and I use 3 Cat5e 100ft cords, over a 100Mbit LAN Router, to go to the other 3 rooms. I plan to upgrade to Gigabit Ethernet but does anyone know if i could get something even way faster than that? Can everyone POST their highest transfer rate between 2 PCs in their home network. And if its really high and you're proud of it, please share with us how you did and what you are using. Be simple and specific. Thanks alot for any input!

Note: I am not complaining about a faulty network cause everything works fine, but i'm limited to 100Mbits and i want something faster.
May 30, 2006 5:52:04 PM

A 100mbps lan connection will have a max of 12.5 Meg/sec transfer rate. But depending on you router you may have less. A gig lan max is 125Meg/sec, but depending on you HD you will get less. Most 7200 rpm drive are around 50-60 meg/sec. If you have 5400rpm drive its around 25-30.
May 31, 2006 1:16:50 AM

lol. i was afraid someone would do what you just did dude. its my fault for having a Title like the one i have. everything you just posted i know but i wanted to know how fast people have gone. and what they did to get that speed. also what hardware. if u have gone faster than me. let me know.
Related resources
May 31, 2006 2:45:22 AM

OK,

With my OLD Old did I say OLD SMC7008ABR max was 4 Meg/sec to my FreeNAS Box. Replaced the Router with a Netgear FVS328 and it jumped to 8.5 Meg/sec. All drag and drop test using a 680 meg iso file. Have a Netgear FVS338 on the way should max out around 12 meg/sec. And 1.8 meg/sec over my 11g AP (Laptop) to FreeNAS.
May 31, 2006 4:36:22 AM

outstanding!!! thanks. but i'm a lil faster, lol. no offense. but i'm looking for a super fast solution cause my bros and I are multimedia freaks and we always exchange all sorts of files. we are musicians, DJs, we make videos, download movies, we make multi-GIG WAV remixes for our parties, multi-GIG tracks for our band, play LAN games all the time, share our work documents from our company, we even mess with 3D animations from time to time. We are talking GIGs galore here, lol. Recently one of my bros mentioned he wanted my whole My Documents folder so that he doesn't have to borrow my computer. Thats about 100GB alone! He said whenever I get a change there was no rush. So hopefully people can see what I want and can suggest something. I did hear about something called 10-Gig Ethernet that came out recently. But any ideas?
June 1, 2006 10:59:52 PM

The peak transfer rates I've seen on my network are around 45Mbps.
My desktop comp has 3 10krpm raptors in a raid0, 3gb of ram, and a marvell pcie gigabit network card. My file server has a 4 disk raid5 on a 64bit 3ware controller (which is in a 64bit slot), 1.7gb of ram, and a 64bit 3com gigabit server nic. My switch is a Netgear GS116 16port gigabit switch. I am currently not using jumbo frames.
I could probably go higher but the raid controller in my server is gettin a little old. It is a 3ware 7410 which is only a 33mhz 64bit card.
June 2, 2006 5:35:10 PM

You could take one of the drives from one computer, install it in the other, and measure file transfer speed among the drives. This will be the upper limit for networking speed. If you get 30 MB/s transfer speed, which would be typical, then it'd be a waste to spend more on high-end networking hardware. Cheap gigabit would serve you just as well.

Of course you can get into high end RAID arrays / etc., alternatively, and these will get you better performance, but you're still going to have a hard time saturating gigabit.
June 3, 2006 4:13:45 AM

how do i go about measuring xfer rates? once you let me know, ill post my results. using winxp
June 3, 2006 4:57:37 AM

hmm. well i used to just drag and drop files from one computer to the next using the usual WinXP Home networking. but that wasn't stable at all. not sure why. but i got tired of it and just downloaded AOL Instant Messenger and it lets me send files or folders to anybody I please. so when i want to send a file to one of my bros, i just have him log onto AIM and i'll send him what he wants. but u could try a FTP Client or some Network Management program. i found a few for free from Download.com but never used any of them. i hear they are great too. other than that. not sure. :( 
June 3, 2006 12:17:53 PM

You should use very large files to measure maximum sustained transfer rates and hopefully avoid file caching effects (by using files much larger than your available RAM). In Windows, I use xcopy in a command prompt. You might combine that with a downloadable stopwatch. E.g. http://www.jfitz.com/dos/index.html

I think a DVD ISO or set of files for a DVD would be a good file size.

Transfer Rate = Size / Total Time

Some sample transfer times for 4.5 GB size:

At 10 MB/s: 7m 30s
At 20 MB/s: 3m 45s
At 30 MB/s: 2m 30s
At 40 MB/s: 1m 53s
At 50 MB/s: 1m 30s
At 60 MB/s: 1m 15s
At 70 MB/s: 1m 4s
At 80 MB/s: 0m 56s
At 90 MB/s: 0m 50s
At 100 MB/s: 0m 45s

You can observe the transfer rate using various tools including PerfMon and Task Manager Networking in XP, but it's best to use an overall figure instead of a momentary figure -- they vary a lot.

Note that transfer rates can be directional read vs. write, and pull vs. push. Push tends to be faster IME. Meaning if you're copying from machine A to machine B, it tends to be faster if you're logged on and doing the operation from machine A.
June 3, 2006 4:31:58 PM

dude are those transfer times Real World situations? i calculated each one and they are perfect, in real world situations those times would be mixed and match. but i do see it says (sample transfer times) so i'm guessing they are not. what i want is your average sustained MB/s transfer rate from 1 computer in your home network, to another computer using a huge file. i don't think you did 100MB/s using a router, switch, or hub. help me out here???
June 3, 2006 4:38:58 PM

Quote:
The peak transfer rates I've seen on my network are around 45Mbps.
My desktop comp has 3 10krpm raptors in a raid0, 3gb of ram, and a marvell pcie gigabit network card. My file server has a 4 disk raid5 on a 64bit 3ware controller (which is in a 64bit slot), 1.7gb of ram, and a 64bit 3com gigabit server nic. My switch is a Netgear GS116 16port gigabit switch. I am currently not using jumbo frames.
I could probably go higher but the raid controller in my server is gettin a little old. It is a 3ware 7410 which is only a 33mhz 64bit card.


Whoa dude thats nice. So do you think people have gone faster than you? lol. yes i'm a speed demon. but your rate is super and the best i've heard of. also do you know if a PCI-Express ethernet solutions are actually faster than regular PCI solutions?
June 3, 2006 4:41:12 PM

Uh, no, I don't get every transfer rate from 10 MB/s to 100 MB/s by 10 -- that's a table I calculated to show how long it would take to transfer a 4.5 GB file at various transfer rates.
June 3, 2006 4:57:44 PM

Quote:
do you know if a PCI-Express ethernet solutions are actually faster than regular PCI solutions?


PCIe can be a bit faster than PCI, but this usually doesn't matter, because you're typically limited by the file transfer performance of your hard drives.

I've benchmarked over 80-90 MB/s RAW NETWORKING bandwidth over PCI network adapters. Caps to emphasize that this wasn't file transfer performance.

If you have an add-on drive / RAID controller on the PCI bus together with a network card, or even a built-in network that goes through the PCI bus as it often does, then you're in a situation where the two parts might be stealing bandwidth from each other, limiting your overall throughput. But I haven't tested this out to see the magnitude of impact. I'd try to avoid this situation by design.
June 4, 2006 1:37:08 AM

Lot of people have gone faster... Me no, but I do regulary transfer 3-15 Gb files at work over 100 mbit ethernet, Patience is a virture.

2003 internet speed record

2005 internet speed record

100+ Gigabit, sorry its a few years old

Just a quick search most are dated; want more, find your own...

Mind you these are not LAN speeds, they are WAN speeds.

10 Gigabit ethernet is available, It will cost you a pretty penny.
Want to play, you have to pay... :twisted:

Better have an open PCI-X or PCI-E 4x slot open, and a fast drive array, or maybe a large ram disk.

Have fun.
June 4, 2006 5:08:04 AM

Quote:

Mind you these are not LAN speeds, they are WAN speeds.

10 Gigabit ethernet is available, It will cost you a pretty penny.
Want to play, you have to pay... :twisted:


This thread is supposed to be about home LAN's, and about file transfers, not WAN transfers to RAM. Show us your home LAN setup that can exceed gigabit, and don't bother if you have hired help maintaining your home LAN.
June 4, 2006 1:31:43 PM

The original question was:
Quote:
I plan to upgrade to Gigabit Ethernet but does anyone know if i could get something even way faster than that?


The 10 Gbit adapters I linked are not WAN adapters, RAM transfers for Internet2 speed records, maybe. This would be one option for something 'way' faster than gigabit. Who knows, some people spend $5k on a gaming box, If this guy wants to spend it on networking hardware and drive arrays to feed it, let him. Anybody have more suggestions, how about Gigabit teaming or something?

A 3 to 5 years ago most would have said Gigabit was not for home networking. I don't know where RAID cards necessary fit into home networking.

I don't have someone maintaining my network. My speed would be lousy, my laptop drive would be the bottleneck. My setup is very humble:

Cable modem-->COTS router-->
--> Dual boot 2K/Linux box (P3-500)
--> Infrant ReadyNAS
--> OKI color Laser Printer
--> LAN of a Wireless router (i.e. as an access point) --> Pair of Laptops

I have gotten line speed out of my 100 mbit network.

In days gone by, the rule of thumb was 1 mhz of CPU / mbit of networking. This was to prevent the CPU from being the bottleneck.
June 8, 2006 7:46:50 AM

I'd also like to see more results, better or worse. Particularly telim's for teamed network benchmarks without considering drive performance.

Here are some of my recent ones in the meanwhile:

Using a single 8.4 GB file (8,425,053,778 Bytes):

3x Raid 0 ---gigabit---> 4x Raid 0: t=108s => 78.0 MB/s
4x Raid 0 ---gigabit---> 3x Raid 0: t=112s => 75.2 MB/s

4x Raid 0 ---gigabit---> 5x Raid 5: t=131s => 64.3 MB/s
5x Raid 5 ---gigabit---> 4x Raid 0: t=118s => 71.4 MB/s

5x Raid 5 ---gigabit---> 3x Raid 0: t=109s => 77.3 MB/s
3x Raid 0 ---gigabit---> 5x Raid 5: t=137s => 61.5 MB/s

These were 3 separate machines, the 4x Raid 0 and 5x Raid 5 are almost full (and the test file is new). The 3x Raid 0 is about 1/2 full.

There is variability in results; better ones were selected above. Push is also faster than pull, so I reported the pushes. I should really report the pulls, and report in terms of read vs write, but bigger numbers are more fun for now.

Please don't try to read too much into the figures -- regard them as simply demonstrative. It'd be incorrect for example to conclude that the 3x Raid 0 is faster for writing that 4x Raid 0 - this result could be due to the fact that the 4x Raid 0 is much more full than the 3x Raid 0.

They are, however, real numbers, measured and calculated accurately as far as I know.

The worst result in these tests was a pull that took 163s => 51.7 MB/s. I didn't test all the variations of pulls.

Both Raid 0 are nVidia RAID. Raid 5 is Broadcom. Drives are similar, 300 GB SATA.

All of the above were with Windows using built-in gigabit networking (2 nVidia, 1 Broadcom) going through a desktop gigabit switch without jumbo frames enabled. Max RAM on any machine was 2 GB.
June 8, 2006 9:06:33 AM

Dear Mr. Madwand!, I am officially in love with you, lol. Thanks dude, I really appreciate this info. I was so dumbstruck to not realize that the PC, which i'm gonna be sending data to, must have a good Hard Drive setup too right? Cause I know I need a good Hard Drive setup so that the network won't bottle neck, but in return, the other side must have a fast setup too? From the looks of it, I see that is correct. Well here is how I see it. In order to instantly improve my transfer rates this is my plan.
1. Buy a Gb Ethernet Router or Switch. Probably 1 of the D-Link DGL's.
2. Buy a Raptor for fast Read/Write. Maybe get another later for Raid.
3. Tell all my bros to buy a Raptor or 2 for less bottlenecks.
4. Make sure all PC's have Gb Ethernet Ports.
5. Ask you guys for any suggestions. :)  If any. Including buying advice.
June 8, 2006 9:28:13 AM

Thanks for the links dude. Those articles were amazing 8O . Where have I been the past few years? anyway, I would never understand how those 10Gb Ethernet Cards are 10x as much as buying 10x1Gb cards, but yeah that was interesting. What I'm hoping to achieve is a speed thats fair, and not draging and hair pulling as the one I have now. How would you feel if you had to wait almost an Hour for a measley 10GB file transfer to finish before you could continue to do anything else. Yes, both systems were slow during the transfer. Will be upgrading soon along with new Network.
June 8, 2006 4:10:45 PM

You need good HD setups on both ends to get the best file copy performance, and also for some local processing, but it's not so important to have RAID arrays everywhere if you're basically reading files from within a program that fit in available RAM and not saving them locally.

Photoshop for example might like nice scratch / cache / swap disks, but if you're viewing big files across the network, and not editing them much, then your local HD performance won't make much of a difference.

Also, I've find that one fast HD setup in either side can help your overall throughput.

So although you can gain from fast HD on both ends, the most important thing to understand is that one side is going to be the definite bottleneck -- 10 GbE ain't going to make a single IDE HD send data noticeably faster than 1 GbE (at least not when that file is not cached in RAM), and the next thing to understand is that a RAID array on one side can improve your file transfer performance, as well as provide redundancy.

Raptors are expensive. Storage Review also likes them a lot for single user performance. But with RAID, HD costs multiply, and I judged it best for myself to pick better capacity / dollar drives. (Strangely, I reported better performance than an above poster with 3x Raptor RAID 0 -- I really don't know why, and it probably doesn't matter, a point is that there's much more here than one best drive type.)

BTW, my switch was a D-Link DGS-1008D. The DGL-4300 doesn't support jumbo frames, but I've never been able to use them anyways -- you need all NIC's to support them, and most consumer NIC's don't. The DGL-4300 gave very similar performance to the DGS-1008D when I compared the two.

I'm still looking forward to telim's reports on his favorite Linksys.

Good luck with your performance improvements -- we're lucky to have so many affordable options these days, although the choices can be bewildering.
June 8, 2006 11:05:10 PM

An hour for 10 Gb is pretty slow. Even for 100Mbit ethernet. That is 2.8 Megabytes/sec or 22.8 megabit/sec.

It is not unusual for me to have to wait 20-30 minutes to copy 5 Gb down at work. Pretty sad when the simulation runs in 90 minutes on the remote system. It's not hard for me to imagine waiting an hour. Luckly my datasets aren't usually over 5 GB.

If you could get 80% of gigabit ethernet throughput then your transfer would take ~1-1/2 minutes. Not a bad improvement from an hour. The hard drive would definitely be the bottleneck.
June 9, 2006 4:39:54 AM

great stuff. yeah i'll just get the new Raptor 74gb with 16MB Cache. The DGL Router. Then do the same for all machines. That should drop waiting periods by half at least. thanks alot.
June 9, 2006 4:47:00 AM

Wow so you know how it feels. lol. Well after my upgrade I tend to upgrade my ISP as well. Don't mean to change the topic on this thread, but i guess i have the rights since its mine. :D  . But I hear Verizon's FiOS is coming closer to home users and it works like a charm. Read it here. http://www.pcmag.com/article2/0,1895,1972495,00.asp
June 13, 2006 10:46:25 AM

First things first:

1GbE ethernet = 128 MB/s THEORETICAL throughput (you'll never realise its full potential), this has to do with networking overhead, windows drag and drop file transfers you'll lose between 35%-40% bandwidth. Linux Samba to windows file transfers 30%-35%.

Another thing to think about the nVIdia 590 chipset can do ethernet bridging, which in effect will double your bandwidth potential, something to think about . . . havent had the chance to play with it myself, but I will soon VERY soon (mwahahaha)

Another thing to think about, is using a SATA port multiplier. Its an added cost yeah, and would require a SIL 3132 controller either on your motherboard, or have an expansion card (i would reccomend PCIE . . .). Anyhow what these devices do, is make it possible to hook 5 SATA drives into ONE SATA port, and when working together with a SIL 3132 controller, they can be run in any RAID configuration, and if your HDDs are fast enough, they will potentialy saturate a SATAII connnection.

Personally, I wouldnt waste my time, or money on raptors, 4x RAID 0, or 5x RAID 5 Seagate barracudas with command queueing would offer more storage, less cost, and perform pretty close IMO. I mean come on . . .you're not going to beable to use all that over the network anyhow, so whats the point . . . 80-90 MB/s is the MOST you'll ever see acrossed the network (if you're lucky). <--- using 1 GbE, that is.

[EDIT]

Another thing I almost forgot about, is the HDD controllers, and ethernet controllers all operate on the same bus, now in the older PCI bus systems, this would have meant you would have had a total of 120 MB/s (ish) throughput TOTAL, system wide. On a PCIE bus it works simular, but with higher bandwidth capabilities. Its something along the lines of 128MB/s per PCIE lane (asyncronous, I think), the thing is, I'm unsure how many lanes are availible for these devices, I would assume something like 3-4 lanes, atleast, but thats just a guess.
June 13, 2006 11:05:13 AM

thats true.I really wouldn't want 4 or 5 hard drives in my case. heck no, lol. I want 2 at the most, and thats good enough speed considering those 80-90MB/s i'll see. You're right about not being able to see the full potential. And thats why a raptor fits the bill. It has a good Average read/write of about 65MB/s. And with 2 of them, thats all the bandwidth i need. thanks for your input.
June 13, 2006 4:12:30 PM

Quote:

the nVIdia 590 chipset can do ethernet bridging, which in effect will double your bandwidth potential, something to think about . . . havent had the chance to play with it myself, but I will soon VERY soon (mwahahaha)


From what I've read, this is uni-directional, outbound only. This means that it's good for a multi-user file server (etc.), where two data streams go to two different computers at the same time with up to 1 Gb/s each, but you won't ever exceed 1 Gb/s to a single computer, even if this computer has 2 nForce 5 NICs.

And of course, to do this sort of streaming performance simultaneously to two different clients, you'd either need the data in cache, or a really nice HD array.

The "proper" way to do this sort of thing is with the 802.3ad standard, and this requires a fancy switch in general, but haven't seen any reports of actual success with this yet, where "actual success" means > 1 Gb/s data transfer performance.

But don't let me discourage you -- good luck, and please report your findings.
June 15, 2006 11:42:44 PM

You dont need a fancy switch at all, you could do Linux ethernet channel bonding . . .of course all your system OS's would have to be Linux . . . with support compiled in . . .

And of course IF the only one machine used 2x GbE bonded, its all going to be outbound (from that machine), but if you use two or more machines with 2x GbE connections bonded, I'm sure it would be bi-directional.

I still advise using eSATA or Mini SAS, the latter of which is probably very expencive. However, I know for a fact that eSATA with a port multiplier is reasonably cheap. You can buy a motherboard with a SIL 3132 controller, the port multiplier, and up to 5 drives (max is 5 per port multiplier). WIth the port multiplier controller, and the SIL 3132 onboard controller, you can run all your disks in whichever supported RAID configuration you wish (which i believe is RAID 0,1,0+1,5, and 10, but basicly boils down to the SIL 3132 support on he motherboard )

Port multipliers cost about $100USD, and for example, the latest AM2 motherboard with a SIL 3132 controller on board is a little over $200, depending on where you buy it, and most of the cost will be investing into your HDDs. You could even buy a cheap motherboard with no onboard SIL 3132 controller, and get an expansion card with a SIL 3132 on it for around $50USD,but in the end, the cost would probably even out, that and the fact the only SIL 3132 controllers i know of that arent in the server class are PCIE 1x, which would limit your bandwidth somewhat . . .
June 16, 2006 5:34:57 AM

Quote:

if you use two or more machines with 2x GbE connections bonded, I'm sure it would be bi-directional.


Do you intend to test this out and post performance results? Native nVidia and/or Linux?

http://forums.techpowerup.com/showthread.php?t=12193

Quote:
DualNet can only improve the outgoing bandwidth. You will be able to send with up to 2000 MBit/s but incoming traffic will still go over one Ethernet interface, resulting in 1000 MBit/s. This means the best gains will be seen with file serving applications. To be able to pump out 200 Megabytes per second you will need fast harddisks, preferably in RAID.
June 16, 2006 9:46:06 PM

ok guys lets stay on topic. but nice Link by the way. :) .
June 17, 2006 12:14:15 AM

With my U.S. Robotics USR5461 WIFi router, I can get about a sustain rate of 9.5MBps between two wired computers, i.e. Linux and Windows.

EDIT: I just tried this to connect my USR5461 behind my Linksys WRT54GS v3 runs on a DDWRT v23 firmware. I put a Linux on this WRT54GS router and was able to achieve the same throughput of about 9.5MBps to transfer DVD files from my Linux box to my Win2k3 (bhind USR5461) through FTP. In other words, this USR5461 really supports 100Mbps connections on all its ports (WAN/LAN). If I do the other way, having WRT54GS connected behind a USR5461, the transfer rate could only go as high as 4MBps which means WRT54GS WAN port can only handle about 32Mbps connections.
June 17, 2006 2:25:31 AM

thats pretty decent. I go about the same. 10MB is the best i've ever seen on my network. anyone get more than 60?
June 19, 2006 7:09:01 PM

Quote:

if you use two or more machines with 2x GbE connections bonded, I'm sure it would be bi-directional.


Do you intend to test this out and post performance results? Native nVidia and/or Linux?

http://forums.techpowerup.com/showthread.php?t=12193

Quote:
DualNet can only improve the outgoing bandwidth. You will be able to send with up to 2000 MBit/s but incoming traffic will still go over one Ethernet interface, resulting in 1000 MBit/s. This means the best gains will be seen with file serving applications. To be able to pump out 200 Megabytes per second you will need fast harddisks, preferably in RAID.


Even though the point is moot, the OP wanted a server for streaming video throughout his house, <---- er, file sharing which would mean outbound packets from the server is really all he cares about anyhow.

Now, I'd like to know how this 'tech person' you linked to got his information because I see nowhere in the white paper (technical brief) where it says it only supports 2 Gbit out . . .

http://www.nvidia.com/content/nforce5/TB-02499-001_v01_DualNet.pdf

It doesnt make sense for ANYONE to cripple a technology by only implementing half of its capabilities, atleast, not in this case it doesnt. This is the problem with information given out over the internet, it is often mis-information. Also note, what you posted for a link is not some reputable website, but rather a link to another forum, which leads me to believe the information is dubious.

In order for me to test what I think is correct, I would need two AM2 motherboards, which at the moment I do not have, but will be getting semi soon.

Now for the HDD side, IF you want to saturate a 2 Gbit connection, you're most likely going to want 4x RAID 0 to do so (atleast) IF you dont care about saturating the 2 Gbit line, then 2x RAID 0 should work fine.

To be honest, I cant see anyone having enough people living in thier house to merrit using more than a single decent HDD in thier home 'server'. However, if the data is 'mission critical', such as work files etc, then perhaps a 3x RAID 5 array is in order ?

Now, to the OP, a Raptor isnt much better than a Seagate barracuda in such a case. I personally have 2x Seagate barracuda 40 GB ATA 100 drives that give me around 50,000KB /s throughput each. Roughly, thats only 10MB/s less, than your Raptor. Last time I benched my SATA 150 250 GB baracuda, it was getting 58,000 KB/s throughput . . . starting to see my point ? Basicly it boils down to you're paying more, for half the storage, and only gaining a marginal boost in improvement. Although, I must admit, I have not been a big fan of WD drives since the 90's . . .
June 20, 2006 12:34:24 AM

very nicely put. well you are right about the throughput, but why is that everytime I talk about a Raptor, 99 people praise it, and then there is always that 1 guy who shames it. lol. but i do have one concern. Do you mean throughput as in Average Read/Write Transfer performance or as in Interface Performance cause there is a huge difference. get back to me on that one dude.
June 20, 2006 2:08:38 PM

Benchmark screenshot I think this says it all, a Raptor has half the access time, but only 7MB/s more throughput (averaged). Seriously, if you're going to be using only two drives, you dont need 'enterprise' drives in your system. Also, I'd check the warranty of the Raptors vs seagate drives. Seagate drives all (mostly, some, such as 'white label drives' dont) have a 5 year warranty, and Seagate has very good customer relations. Since WD has been around for awhile, I would imagine they couldnt be terrible in this arena, but the last time I checked thier warranty was still lacking by comparrison. However, unless I'm mistaken, the Raptor is classed as an enterprise drive, and may have a longer warranty.

I cant stress enough that you DO NOT NEED, that extra 7MB/s for the added cost . . . use that money elsewhere in your system, such as your video card, or something :) 

Let me put it another way you may understand more clearly what I mean. IF you had been buying Seagate barracudas for a generation or two, then suddenly Raptors appeared, and you HAD to have it, and you bought one, I think you would be atleast midly dissapointed. The meaning here, is that on a desktop PC, where you're only doing occational filesharing (1-2 times a day or so) you will notice very little if any perceivable difference. The only time you'd know the difference is when, and if you benchmarked it. Even then, the difference is basicly only for bragging rights, if anything. Now, if this was for a small businesswith 10-20 people ( or slightly more ), using these in some kind of RAID configuration, then, you would notice a perceivable difference, especially if the disks were being accessed often.

http://www.newegg.com/Product/Product.asp?Item=N82E16822148140
Quote:


Pros: Very Fast, Very Quiet, Very Cool. Good price for huge amount of storage....


Cons: Nothing at all so far.... More ยป


Other Thoughts: I have two of these in RAID 0 in 3.0gb/s configuration. It only took like an hour and five minutes to format this behemoth 600g array. I have an older seagate 250g IDE drive, and as i remember it took over two hours to format that. I work with very large video files, and let me tell you, these can move data very quickly, especially striped in 3g/s. Oh and dont forget to remove that nearly microscopic jumper to activate 3g/s.



I didnt read any of the reviews on the 150GB Raptor . . .but man . . . $269USD, and yeah, they are one of the few WD drives that have a 5 year warranty (1.2 million hours MTBF, mean time before failure . .).


Ah yeah, time for another edit. I was looking at stuff on newegg the other day, and was noticing a netgear 1GbE switch that was on sale, and it had it listed in its specs, that it supported 2GBit/s throughput per port. I dont know if you understand the ramifications here, but from what experience I have
in this area (believe me, its more than just a little), this had only been previously done on cisco routers/switches, and some other expencive enterprise switches that support load balancing. Basicly, what I'm getting at here, is that you COULD go with a nVidia 570/ 590 motherboard, tie both ports to the switch, and run 2GbE out on a single line. In order for this to work for every PC in the house however, you would need either a switch in each room, OR one switch in some room, with two cables for each PC connecting to it, and of course, the obvious nVidia 570+ chipset motherboard for each PC. Quite the interresting time for technology we are living in now :) 
June 20, 2006 7:40:22 PM

wow, what an outstanding point man. you see by the time i read your post i was already convinced that i didn't need a raptor, and was ready to invest on the delicious looking Seagate. But when i saw the prices between the seagate and raptor 74gb, i still have a tough decision to make. considering the new model of the 74GB Raptor is way faster than the previous model. and with the rebate offering, we are talking only a few bucks difference. the only thing i can do is wait for conroe to come out and then worry about buying a good fast drive at that time.

thanks dude, very persuasive indeed. lol
June 20, 2006 8:39:58 PM

Quote:

Even though the point is moot


You brought it up here, btw.

Quote:

It doesnt make sense for ANYONE to cripple a technology by only implementing half of its capabilities, atleast, not in this case it doesnt. This is the problem with information given out over the internet, it is often mis-information. Also note, what you posted for a link is not some reputable website, but rather a link to another forum, which leads me to believe the information is dubious.


That post was from a site admin/editor/reviewer/whatever. Doesn't make it 100% reliable, but what do we have to compare that with now? Personal high hopes for the technology? If you look at this question seriously, and not just with enthusiasm, you'll see that both possibilities exist, and remain to be verified.

Quote:

In order for me to test what I think is correct, I would need two AM2 motherboards, which at the moment I do not have, but will be getting semi soon.


Exactly -- jumping into this technology is not cheap. If you're laying out this much money, it'd be pretty smart to do so after seeing some successful results posted somewhere. Otherwise, what you're doing is buying based on marketing, exactly what the marketing folks want, not what the engineering folks would advise.

Quote:

To be honest, I cant see anyone having enough people living in thier house to merrit using more than a single decent HDD in thier home 'server'. However, if the data is 'mission critical', such as work files etc, then perhaps a 3x RAID 5 array is in order ?


Fact is that tons of home users have RAID arrays and multiple 100's of GB of personal data, some of it redundant and fully backed up, etc., and that data can grow fairly rapidly. Whether or not that's 3% or 1% or less of the population whatever is of no consequence when you have such data.

That said, there's much much to be said for the simplicity of non-RAID'd single drives. If you don't really need their features, then you'd probably be better off avoiding them.

Getting back to the semi-off-topic teaming, I contacted nVIDIA, and they said that DualNet teaming supports 802.3ad. I'm very happy to hear that, but would still recommend holding back and waiting for actual results before spending a lot of money and hope on this.

So who's going to post such results for me? :)  If you do, you'll probably win this little competition of "highest MB/s transfer rate on a home LAN", for some time at least...
June 20, 2006 9:15:36 PM

yes sir, but it seems to me that no one really cares about this post. so far only a few loyal posters left a transfer rate if at all to help with my mission. i wish more people would just go home, send a file to another PC on their LAN, then come back and say "this is what i'm getting or got" "and this is the setup i'm using to achieve this rate". Oh boy would that be a perfect and plain simple world. but oh well, i just hope every 20 posts has 1 transfer rate. i'll never know.
June 21, 2006 7:23:35 AM

Madwand: READ the friggen white paper link i posted. Also, do you even know what ethernet bridging is ? I'm assuming no, because if you knew what it was, and you read the technical brief, you would know its capabilities already . . .
ethernet channel bridging ISNT somethign nVidia invented, its been around for years, but because of either the cost, or the OS / hardware involved, it wasnt feasable for use in a home network envoirnment (unless that home used Linux only) Now, since nVidia has seemingly implemented it in hardware, the need for proiprietary switches, or the use of Linux is irrellivent. The end result is that you, me and every other PC owner who uses WIndows as thier primary OS, can enjoy ethernet bridging. Now, quite being so adveserial.

Trance, the Raptors ARE NOT way faster . . . they are 'marginaly' faster if, and thats compared to my 1 + year old SATA I drives. If You consider the added cost, and the performance difference, its no where near worth it. Unless of course, you where a buisness, needing the miniscule performance gain, and could write it off. In either case, IF you were a business, and used solely SATA I, I'd have to question your IT skills.
June 21, 2006 7:50:26 AM

Trance, if you tell us how much money you're preparred to spend on this, we could better tell what the path best for YOU. There are a few different possibilities.

If I were to go by my needs, and what I'd spend in such a situation, I'd buy cheaper larger drives capable of substaining close to that of the Raptors, and spend the additional cash saved, on 1GbE cards, or motherboards with integrated 1GbE ports, and be happy with that.

Sure, you can spend hundreds on HDDs that wont even benifit you much in the long run, and you can even spends hundreds, if not thousands on the network, to keep everything maxed.
June 21, 2006 8:08:58 AM

Yyrkoon, all I asked for was some caution and testing before jumping on the gung-ho buy bandwagon. You've made this adversarial. I won't even point out the several mistakes in your post beyond their simple existence.
June 21, 2006 7:00:01 PM

well here it is, i plan to spend about 300-400 USA dollars to upgrade my setup. that includes a router or switch, and hard drives, i have the cables and have just made all our PCs 1Gbit capable. Anymore and we're probably talking business corps and servers.

I see the Raptor 74GB is going for about $130 with rebates. the seagate looks to be $120. very tough. but i couldn't find the exact model you are using. I like the 58MB average throughput and if i'm guaranteed that performance then i'm off to get the seagate. but like i said, i can't find ur model, how do i know a slightly different model will give me the same if not better performance. it all comes down to benchmarks for me, and if i'm lucky, i'll find a good seagate for under 200 bucks with the same performance. Look at Tom's Chart. I see your point.

http://www23.tomshardware.com/storage.html?modelx=33&mo...
June 22, 2006 4:15:09 AM

1st question. Router or switch? How's your current router holding up as a router? Do you forsee any need to upgrade the router itself or add a wireless? If no, then just get a switch. If yes, then still consider a switch; consider a router + separate switch. A separate switch will give you additional ports and a bit more cabling flexibility, and some optional features like jumbo frames.

2nd question: Which HD? Note that average transfer rates can be very misleading -- they're measured over the entire drive, and when you compare a very big drive to a small drive, the really fair measurement is the performance of the big drive over the range of the small drive. Moreover, in this range, the average seek performance of the big drive will also be better, because it doesn't have as far to go. 10K vs 7.2K is a significant difference, but so is the 74 GB vs. 300+ GB.

Note also that the capacity difference helps you in other ways -- you're less likely to heavily fragment the drive, and you can even use the rest of the space for secondary purposes such as a backup. I have a 300 GB PATA drive just for the OS. A couple of OS's actually. Still way overkill. So I have a big >200 GB partition that I just use for backups.

StorageReview.com has the best info for this sort of decision AFAIK -- I recommend checking them out.
June 22, 2006 5:05:01 AM

ok this is good. I have an old Microsoft Router 10/100Mbit. But i want something that will give me wireless, internet for all the PCs in the house, and good packet handling for when we play LAN games, which we do. Can a switch do all that? I know the D-Link Router can. If so, then let me know and i'll buy that.

About the hard drive. Just as I was sold on buying one of the 250GB that Read/Write near as good as a Raptor. I run into another specification which is "I/O performance". And I think from what I see, that has alot to do with how fast your games can load maps and launch Apps and Boot OS. So help me with this cause i'm confused? It seems the Raptors outperform all drives when it comes to "I/O". i could use some insight.
June 22, 2006 6:57:08 AM

Quote:
well here it is, i plan to spend about 300-400 USA dollars to upgrade my setup. that includes a router or switch, and hard drives, i have the cables and have just made all our PCs 1Gbit capable. Anymore and we're probably talking business corps and servers.

I see the Raptor 74GB is going for about $130 with rebates. the seagate looks to be $120. very tough. but i couldn't find the exact model you are using. I like the 58MB average throughput and if i'm guaranteed that performance then i'm off to get the seagate. but like i said, i can't find ur model, how do i know a slightly different model will give me the same if not better performance. it all comes down to benchmarks for me, and if i'm lucky, i'll find a good seagate for under 200 bucks with the same performance. Look at Tom's Chart. I see your point.

http://www23.tomshardware.com/storage.html?modelx=33&mo...


Ok, 300-400usd sounds simular to what I'd spend. The drive Listed in my benchmark is around 2 year old tech, I was just using it as an example, the newer SATAII drives will perform close to the same or better. Which, btw doesnt really have anything to do with SATAII, just the platter density and such. Barracudas (SATA drives) have always done native command queuing, so I didnt even bother mentioning that.

Hell for that matter, you dont even need to buy Seagate, you can buy any ole well performaing drive, but I recommend Seagate because of thier customer service, warranty, and high performance / cost ratio. Anyone can buy a 'lemon' product from anyone, and usually customer service is an after thought, but handy to have in a pinch when needed (dont want to be dorking about with buttheads who try to renig on thier 'warranty').

Despite what people would have you beleive, access times arent what the manufactuers would have you beleive. Just for a quick example, 5-6 years ago, I bought an Adaptec 29160 SCSI controller, and an Ultra plex UW Plextor CD reader, for the speed, and even after re-encoding severa Music CDs for my awsome creative MP3 player (with a whopping 32 MB memory! !), the feeling I was left with compared to my regular IDE reader, was simular to eating good, but stale food. meaning yeah, It was faster, MAYBE even a good deal faster (20% or slightly more), but real world performance was like 30 seconds . . .500 usd for a 30 second difference . . .Point is, I can link you to the fastest HDD solution on the planet, but will you use all that power ? will you even NEED it ? I think not.

My suggestion would be to buy one or two decent performaing, drives, on sale with rebate if possible (i will STILL avocate Seagate over WD any day of the week, I've had nothing but good solid performance from them).

Find a good switch for a decent price, maybe even something like the Netgear I saw last week with 2GbE support per port (never know if you want to do ethernet bridging in the future)

As for NICs, the last time I researched this more than a year ago the INtel PCI NIC in this
'roundup' was the best, it is possible it still is. Anyhow, yo ucould spend the remaining of you budget on these, since 400 usd is hardly enough to upgrade 4 or so systems to AM2.

Here, you'll save a bit of money while getting more HDD space, making your HDD selection a bit more future proof, the netgear GbE switch WAS on sale for 69 usd, and offers some good options for possible future upgrades. Keep in mind I did not research this piece of hardware past the several reviews I read on newegg, which can be a good indicator, but isnt always fool proof. The majority of your money would be going into your NICs, which if you're more concerned about bandwidth than anything else, it would be a good investment. Last time i Priced these around a year ago, they ran about 60USD each, hyopefully they are a bit cheaper now, since they would eat up a pretty good bit of your biget by themselves . .

Now, in case ANYONE got the wrong idea Raptors are not a bad drive, they are the highest performing drive on the market in the SATA arena (as far as I know), however it is my opinion, that these drives are meant for people who buy stuff like FX-62 CPUs, and have more money than they know what to do with, yeah sure they are the 'best', but are they worth the extra cost, and lost storage space ? I dont think so. Irreguardless, you're budget doesnt support the idea of buying them IMO anyhow . . .

[EDIT]

Sorry for all the typos, Im too lazy to fix them atm, as ive been out all day installing a T1, and im exausted.
June 22, 2006 7:26:39 AM

Quote:
1st question. Router or switch? How's your current router holding up as a router? Do you forsee any need to upgrade the router itself or add a wireless? If no, then just get a switch. If yes, then still consider a switch; consider a router + separate switch. A separate switch will give you additional ports and a bit more cabling flexibility, and some optional features like jumbo frames.

2nd question: Which HD? Note that average transfer rates can be very misleading -- they're measured over the entire drive, and when you compare a very big drive to a small drive, the really fair measurement is the performance of the big drive over the range of the small drive. Moreover, in this range, the average seek performance of the big drive will also be better, because it doesn't have as far to go. 10K vs 7.2K is a significant difference, but so is the 74 GB vs. 300+ GB.

Note also that the capacity difference helps you in other ways -- you're less likely to heavily fragment the drive, and you can even use the rest of the space for secondary purposes such as a backup. I have a 300 GB PATA drive just for the OS. A couple of OS's actually. Still way overkill. So I have a big >200 GB partition that I just use for backups.

StorageReview.com has the best info for this sort of decision AFAIK -- I recommend checking them out.


you know of a switch or router that supports jumbo frames, and is decent, that wont blow his budget ? I dont, that doesnt mean they dont exist, how about linking some parts . .

about the HD speeds, yea platter location, and if you go by the same rate the Raptor was dropping towards the end of its platter, and made it 300GB, it looks to me as though its perfomrance would drop significantly (as in ZERO MB/s). Also have to keep in mind that a drive more than 50% larger than a raptor, and only trailing by 7MB/s is pretty dahmed good. A Raptor would be a good option for a pure OS Drive sure, IF you can afford it.

As stated in my earlier post seek time has nothing to do with real world performance, well it does, but like it said, the difference WILL leave a stale taste in your mouth IF you care anything about your money, and how you spend it.

As for transfer speeds, the worst possible benchmark you can do on a HDD is a combined read/write benchmark where it reads data then writes it back as fast as it can, I thought this benchmark I have now had this, but seems it doesnt. This is the type of test I preffer when benchmarking a drive, but since not availible, I simply did with what I have., Ive already proven atleast to myself, that seek times are even more misleading than transfer rates. BESIDES, the newer Seagate drives i linked to the other day, have the same, or closer seek times to the raptors :)  Less cost, nearly triple the storage space.

You also forgot to mention that while RPM speed can play a big factor in performance, it can ALSO play a big factor in system heat, and running anything over 7200 RPM with out some sort of active cooling (such as a fan /HDD mount, fan blowing dirrectly on the HDD itsself ) isnt very smart.
June 22, 2006 7:31:53 AM

Trance: AH yeah before i hit the rack, WE forgot a very important factor, as in the exact specs for you current system NOW. If' you're currently using a dated PCI system, all this could be moot, since the max bus throughput would only be 120 (ish) MB/s anyhow, but I've been assuming you're using a decent PCIE system . . . Also, knowing exactly what you have now could save some additional monies, if you have some decent parts already (HDD wise)
June 22, 2006 9:20:23 AM

Quote:
I have an old Microsoft Router 10/100Mbit. But i want something that will give me wireless, internet for all the PCs in the house, and good packet handling for when we play LAN games, which we do. Can a switch do all that? I know the D-Link Router can. If so, then let me know and i'll buy that.


If the DGL-4300 has enough ports for you, and you don't need jumbo frame support, and standard medium-range 802.11g wireless is enough, then it's a good choice.

You could go a bit cheaper and get more gigabit ports by not upgrading the router for now and just getting a desktop gigabit switch (e.g. D-Link DGS-1008D), and you can even consider buying a separate wireless router + gigabit switch which could cost less than the DGL-4300 together, and in some cases could give better wireless range.

(The future of wireless is 802.11n, but unfortunately, the pre-n products aren't really ready and are relatively expensive. This might be a reason to consider holding off upgrading to wireless if you don't really need it now, and want the best performance when you get it. In my thinking, it's a good time to get gigabit, but not the best time to be adopting wireless because it's changing significantly. But you can get stable 802.11g wireless affordably now if you're not going to be upset with the faster/longer range stuff coming out and becoming stable and cheaper over the rest of the year or so.)
June 22, 2006 7:54:59 PM

Quote:

About the hard drive. Just as I was sold on buying one of the 250GB that Read/Write near as good as a Raptor. I run into another specification which is "I/O performance". And I think from what I see, that has alot to do with how fast your games can load maps and launch Apps and Boot OS. So help me with this cause i'm confused? It seems the Raptors outperform all drives when it comes to "I/O". i could use some insight.


StorageReview has a decent approach, where they try to go after application performance instead of simple synthetic benchmarks. What they do for this is typically like: Create / obtain a general application benchmark. Run that test, and capture all the drive accesses. Then run just all the drive accesses, and measure performance. This is cool, and valid from the pure drive access point of view, but still exaggerates the effect of drive accesses. This is intentional, but what I'm pointing out is that this is still not an ultimately meaningful real world benchmark, where you're measuring overall time. You're just measuring drive access time here. In the real world, you're doing a lot more than just drive access, and because of file caching, etc., the drive access is usually not as big of a part of the overall performance as such extracted benchmarks would lead you to believe. Moreover, with multiple drives, there are tons of ways to configure these things, and results could be very different in cases where the OS is on one drive and the primary apps on another, or everything on one drive, or the temp folder, swap file or Photoshop scratch file on another, etc. So their benchmarks, while conceptually valid to a fair extent, still don't tell you the entire real world impact or of all the alternatives.

Gaming in particular is not really drive-limited. It would be a really bad idea to make a game which was. So it would generally be a big mistake to take video card or RAM dollars and put them into a very fast and expensive hard drive. But yes, in some games which have huge amounts of data to pull of the drive in parts -- zone loading -- then for just that time, you can see significant benefits going from very old/slow/full/fragmented drives to new/fast/empty/unfragmented drives. But realize that this will usually have no impact on FPS for example, and in the general case, you might be talking about 30s vs. 20s for a big zone and an average drive vs. a fast drive.

So the real challenge here is to find out, in your particular case, where the drive accesses are really significant, and then to figure out how you can improve upon them. BTW, often, the best solution turns out to be avoiding the disk altogether, and this is often done by increasing the RAM available to the OS and/or the application.

The Raptor 150 actually tends to have a better GB/$ and performance figures than other Raptors. So if you really want the fastest desktop drive, then that's probably the one to get. (SR has more details, and sometimes the big new drives can do nearly as well - check them for details if you want/need more space.) But you've probably noticed that Raptor 150's aren't cheap. How much is drive performance worth to you? How often are you actually hitting the drive? In what manner? Are you moving a lot of data around from one place to another and accessing a ton of small files, or pulling data in and out from the applications sparingly? In the latter case, a drive setup with good sequential transfer rate could be better / good enough; in the former, and drive with good seek performance could be better.

The Raptor question is pretty much a question of money and capacity. No doubt that it's generally faster. But they don't have much capacity in modern terms. And they're much more expensive than the norm. If capacity and money don't matter to you, they're the ones to get. But with limited budgets, we have to make those trade-offs as suit us best, and it's really not that hard to live with a bigger drive that performs well, but not as well as a Raptor.
June 23, 2006 12:26:29 AM

It still sounds to me as though some one is trying to confuse the situation. There two HDD's, same test, and differences listed, and yet for reasons unbeknownst to me, is STILL unwilling to accept the data.

Provide some links, or some useful data instead of hersay, or just drop it.

Ive already proven TWICE that modern Seagate drives perform admirably well for less cost, and over three times the storage capacity. Having given my own benchmark (irreguardless if its something you like or not, both drives were tested with the same program, which btw,alot of people use now days). I've even given a users review who actually OWNS on of these seagate drives, and a link to the review, and product.

Before you think, I'm being agessive towards you madwand, provide some real data, links, or some usefull information other than hersay. I wont think you're being agressive, hell I may even learn something, but until you prove something to me, I'm going to consider you wrong.
!