SATA cables: Is there a difference?
Written on August 11, 2011 by Matt BachIntroduction
Edit 2/11/2013: Due to the date of this testing and the fact that this question still comes up fairly often, we have decided to revisit this topic with more modern hardware in our SATA 3Gb/s vs. 6Gb/s Cable Performance (Revisited) article. Check it out to see if the results are any different using newer hardware and cables.
In the offical SATA-IO document "Fast Just Got Faster: SATA 6Gb/s", the following is stated:
The same cables and connectors used for current SATA implementations can be used to connect SATA 6Gb/s devices. SATA-IO recommends utilizing quality components to ensure data integrity and robust operation at the fast 6Gb/s transfer rate. Cables already at the threshold of 3Gb/s operating margins may experience lower performance than expected at 6Gb/s due to an increased number of resends.
In other words, as long as you are not using cheap knock-off cables there should not be any difference. We prefer to have empirical data to back up these types of claims however, so we decided to compare a set of supposed SATA 3Gb/s cables against a SATA 6Gb/s cable.
Before running the speed tests, we cut open a set of cables to see if we could find any design differences. The big things to look for at the number of wires and the gauge (size) of the wires.
| Asus SATA 6Gb/s (Black) | Asus SATA 3Gb/s (Red) | Asus SATA 3Gb/s (Black) | Intel SATA 3Gb/s (Blue) |
In the pictures, the sleeved wires are all the signal wires, and the unsleeved are the ground wires. You can see that all the cables have the same gauge of signal wire, although the Asus SATA 6Gb/s and the Intel SATA 3Gb/s both have an additional ground wire. This is not for performance reasons, but simply due to the design of the cable. Instead of having the cable in one piece, those two cables are more like two separate cables fused together. This should not have any effect on performance, which hopefully our speed tests will verify.
Speed Tests
To compare the speeds for each cable, we configured our test hard drive as a secondary storage drive and ran the benchmarks with each separate cable. We used a very fast SSD SATA 6Gb/s drive on a SATA 6Gb/s port, so any loss of speed should quickly become evident. The benchmark we used was CrystalDiskMark.
Test Hardware:
| Motherboard | Asus P8P67 Pro |
| SATA Controller | Intel P67 SATA 6Gb/s |
| RAM | 2x Kingston ValueRAM DDR3-1333 2GB |
| CPU | Intel 2500K Quad Core 3.3GHz |
| Hard Drive | Intel SSD 510 250GB SATA 6GB/s |
| OS | Windows 7 Ultimate 64-bit |
With all that said, lets take a look at the actual performance of each cable. All results are in MB/s.
| Asus SATA 6Gb/s (Black) | Asus SATA 3Gb/s (Red) | Asus SATA 3Gb/s (Black) | Intel SATA 3Gb/s (Blue) | |
| Sequential Read | 464.2 | 464.0 | 462.7 | 460.5 |
| Sequential Write | 326.8 | 328.2 | 328.9 | 327.5 |
| Random Read 512KB | 314.6 | 314.8 | 315.3 | 315.1 |
| Random Write 512KB | 301.4 | 301.2 | 300.8 | 301.3 |
| Random Read 4KB (QD=1) | 19.8 | 19.8 | 19.8 | 19.8 |
| Random Write 4KB (QD=1) | 44.0 | 43.9 | 43.9 | 44.0 |
| Random Read 4KB (QD=32) | 80.3 | 80.1 | 80.2 | 80.1 |
| Random Write 4KB (QD=32) | 50.7 | 50.6 | 50.7 | 50.6 |
From these results, we can pretty conclusively determine that all of these cables are running at the full SATA 6Gb/s speeds. There are of course some minor differences between the results, but they are all well within normal testing variations.
Conclusion
In conclusion, our testing has agreed 100% with SATA-IO's statement that SATA 3Gb/s cables will work perfectly fine with SATA 6Gb/s drives. This wasn't much of a surprise to us, but it is always nice to have hard data backing up a claim.
Great stuff - thanks!
Funny how I was just completing a new build with an ASUS P8Z68-V PRO/GEN3 motherboard and, aside from the 6 GB/s cables included with the Intel 520 Series SSD I installed, I pondered about the black w/black connector head and black w/white connector head SATA cables included with the motherboard--i.e., which set was 3 GB/s rated and which was 6 GB/s rated because neither set were marked nor were they distinguished in the motherboard manual. So, I called ASUS tech support and asked. Guess what?...matching perfectly with the pictures in this article, the ASUS tech said the black w/white connection heads were the 6s and the black w/black were the 3s. The main reason I went to all this trouble to ask is that I wondered which to use to connect up a couple of WD hardrives. Well, well, well...according to this article, it really looks like it doesn't matter which cables you use. it's the DEVICE that determines throughput capability. If the DEVICE is 6 GB/s rated, use whichever cable you like....you'll still get 6 GB/s capability.
shutup, nerd
"...it's the DEVICE that determines throughput capability..."
Point of order: No. A bank of drives in RAID0 have a throughput cap of 56k if accessed over POTS.
Throughput capability is defined by the whole.
lol good one.
Awesome work. Nice pics and some good wire stripping.
Thanks a lot for the info and the testing, it helps a lot :)
I was confuse on this matter. The info was great. You guys did a great job. THANK YOU
It may be a problem when you use some drive which can support more than 500MB per sec since for 3 Gb/s cables just guarantee only 384MB per sec
Unless you are trying to use super cheap cables from Ebay or somewhere like that, you shouldn't see any performance difference no matter the drive you are using. A wire is a wire, and as long as the gauge of the wire is correct, the end connections are good, and the right metal is used, there is no performance difference between one cable and another.
Not hardly....... These SATA and other" paired" wire runs are Hi frequency
transmission lines, with a "characteristic" impedance that is subject to, wire
type (solid or multi- pared twisted core), dielectric type, spacing, and proximity
to other conductive media (ground).. Much like an antenna feed....
i will wreck you
Hay. "Not a nerd", someone has hi-jacked my ID to post that reply,(although it's correct)..
And for that you Reply "i will wreck you" ?????.......You Got to be a Nerd...(pun intended)
Saturate the bus at its full spec' length; see what happens across a broad selection of 3 and 6Gb/s cables. 500MB/s = 4Gb/s. You need to hit >750MB/s to really test a 6Gb/s cable. Tiny differences in insulation thickness and capacitance make a proportionally bigger difference to signal quality the harder you drive the connection; it'll still *work*, but increasing error correction and data re-requests will begin to eat into the maximum bandwidth on a low quality (not necessarily inexpensive) cable.
I'd guess that: If a drive's read capacity is maxed at e.g. 500MB/s and a SATA cable is producing errors that require data to be re-sent, that (very recently read) data will still be in the drive's cache (assuming it *has* cache; apparently some sandforce-based drives don't use it at all: http://www.legitreviews.com/ar....
If the guess above is right, even though the cable is introducing a high error rate, as long as the error rate only slows the cable's data rate to something between 500MB/s and 750MB/s, you won't see any measurable difference at all in throughput from a 500MB/s drive because of the read buffer. You could be re-transmitting half your data and not know it.
I don't know enough about the SATA controller spec to say whether there's an additional buffer between the drive's putative cache and the cable itself that would amplify this effect, or even introduce it to the sandforce controllers mentioned above, but... "probably"?
This feels like a simplification of how the buffer actually affects throughput, but it's in the right ball-park.
Numbers speak.
The latch cables that Asus has been pushing are crap. This test not only proves the new connections are a gimmick but rip off. The latches pop off when you push and pull to remove them. The only purpose for them is to add to their bottom line in motherboards and drives and offer no benefit whatsoever
In our experience, the latches work great to keep the cables seated. This may not be a huge deal for users that build their own computers, but since we ship nearly every system we build they greatly reduce the number of times the cables come unseated in shipping. We build and ship hundreds of systems a month here at Puget Systems, and we have nothing but good things to say about SATA cable latches.
In fact, nearly every manufacturer (not just ASUS) now uses these latches. If you check out our updated SATA Cable article (http://www.pugetsystems.com/la... both ASUS and Intel cables now have these latches. And if you look online you'll find that most other brands (Gigabyte, Zotac, MSI, etc.) use them as well.
XP crap just wasted 25 bucks on two cables to get the same speed ... hahha, damn tom share posts ...
Very kind of you to take the time to prove the theory and then explain it to the rest of us so clearly. Certainly gave me the reassurance I needed. Have just upgraded from a failing SATA II drive to a new SATA III drive without wasting money on a new cable thanks to you. Much appreciated.
Very interesting.
Thank you for taking your time and doing full test!
Actually they are simply SATA cable not SATA I, SATA II or SATA III.
Still suffering from the old PATA 40 vs 80 wire cable syndrome I asked myself the same question regarding SATA cables and could find nothing in the repair manuals or mother board books so I just used the old Asus SATA cables for the new hard drive, with only the best results. Thanks for all the work and the reassurance.
I just got stuck with some terrible "Sata 3" cables off ebay. I was wondering what the hell was going on when I rewired my desktop and after power on immediately got a SMART error on a brand new 730 SSD (Intel stuff works perfect out of the box..always..period), had another SSD disappearing and reappearing, and then one of the SSDs in my RAID array was only running at 3gbs on a 6gbps SATA port which obviously slowed the entire array down to 3gbps * 2 speeds.
I don't know whether it was because I used round cables or because I used fairly long ones (18"), but $2 / cable isn't good quality stuff.
Round cables and 18" cables should both be fine, it was likely just that the brand of cable you received was not a very high quality brand. In fact, $2 per cable is actually more than we usually pay for a SATA cable. The cables we use here when a motherboard does not include enough cables are the $1-2 SATA cables from MonoPrice (http://www.monoprice.com/Produ.... We've never tried their colored cables, but the black ones we have had no problems with even though they are fairly inexpensive.
Thank-you. Well written and concise. Important information! Unlike a lot of the other crap on the web about technology.
This answered my question. I hope nothing has changed in three years. Hate to have to buy all new right angle SATA cables. Thanks for the testing!
Thank you!
Thanks!
Great stuff..thanks.
Cheers mate!
we are factory specializing in the manufacture sata cable and other data cable,power cable,av cable
www.stcdz.com
Technically speaking you are 100% wrong.
1. All cables that you've tested are not truly compliant with 6GB (it requires separate shielding of all 4 data cables in form of SFP+ twiax)
2. All cables that you've tested are over specified 3GB cables - where for 3GB you are not required a run any shielding up to 50cm, cables that you;ve provided here will have a sufficient frequency response due to separation of TX and RX lanes.
3. 6GB cabling warrants through grounded at the plug shielding that any cross talk from neighbouring cables will be lessened, the noise reception on the cable will not cause a chaos on motherboard connector -> sata controller. You completely ignored this fact and did perform a test on a single drive at a time, where's a standard 3GB cabling in setup of multiple disks being hammered with 6GB high IO (lest say raid arraid) - those cabling will prove to be a catastrophic failure point.
4. You run a laughable testing with CrystalDiskMark, on a machine that does not have a ECC ram, you don't run any form of error discovery post your test, so yes you may write and read from you disk at blazing speeds, but it might be just the garbage. You've not performed any s.m.a.r.t. monitoring for error count. You've not performed any test that is very heavy on sata cabling, just maxed out your disk performance through single process.
5. You've not provided a single form of evidence like scope trace "eye diagram", current consumption by controller testing (which may lead to premature capacitor death on your mobo) to prove that there is no distortion in data transfers.
Essentially your test (and previous one as well) is what we laughably call in development as "it worked on my desk for 10 seconds so is production ready".
Please do your home work before you publish phony results that make statements that people that write standards are wrong. Lame.
Tom, technically speaking, you're a shitwad. Essentially, you're the kind of developer that we sysadmins delete tickets from arbitrarily because of the way you treat other people.
Please, grow up before posting on the Internet. Lame.
Interesting, but I would not necessarily expect a speed difference. I would expect potential unreliability that would increase with cable length. I think a more useful test would be to check the frequency of read or write errors, or other bad behavior. Most useful to average people would be to do this with reasonably long cables so that if no reliability concerns show up, the user could then be fairly sure that his or her build would be fine. (Or, is the theory that this would show up as a speed drop due to automatic retries?)