Did a few quick benchmark on Sata3 vs Sata6.

LOCO LAPTOP

LOCO LAPTOP

[H]F Junkie
Joined
May 4, 2006
Messages
12,620
So got my hands on 2 new drives and a new motheboard that support sata 6.
Since I feel nice I wanted to post a few benchmarks to compare. :)

50MB run
50mbnew.png

100MB run
100mbnew.png

500MB run
500mbnew.png

1000MB run
1000mbnew.png


EDIT: 9/27/10
2000mb run
2000mb.png

4000MB run
4000mb.png


All these test were done one at a time while the pc was doing nothing else to provide the best results.
 
Last edited:
The title threw me off a little bit, You are comparing SATA2 drives with SATA3.

SATA2 = 3Gb/s
SATA3 = 6Gb/s
 
I have to redo that sata 6.0GB/s test. Turned out my drive was running at 3.0GB/s! :eek:

EDIT: updated with correct speeds. :)
 
Last edited:
So this basically affirms what most people have said, that SATA 6Gbps does basically nothing for hard drives. I would, however, like very much to see benchmark results, as well as real-world impressions on the difference between SATA 3Gbps and 6Gbps on an appropriate SSD - the only one that seems to have popped up to my knowledge is the C300.

EDIT:

I have to redo that sata 6.0GB/s test. Turned out my drive was running at 3.0GB/s! :eek:
Click to expand...
Lol. I guess I'll reserve judgment then.
 
Mr. Pedantic said:
So this basically affirms what most people have said, that SATA 6Gbps does basically nothing for hard drives. I would, however, like very much to see benchmark results, as well as real-world impressions on the difference between SATA 3Gbps and 6Gbps on an appropriate SSD - the only one that seems to have popped up to my knowledge is the C300.

EDIT:


Lol. I guess I'll reserve judgment then.
Click to expand...

Yeah I was kinda disappointed until noticed that it was running at 3.0 speeds lol. but yes I have now updated the results. ;)
 
Hi Just a side-note

If you compared the results for WD6400AAKS Blue RAID-0 test results and Corsair P128 (Writes Results), you can see that WD 640GB Blue is really a wonderful drive even with only 16MB cache. Thought I am not sure how the new revision performs.
 
Bear in mind that there's more to performance than the interface itself. Implementation and drivers play a major role in performance, and few can match Intel in that department.

Expect to see a larger boost with Sandy Bridge (native SATA-6 support) and the next generation of SATA-6 SSDs.
 
Try increasing the test size to 1000MB+ and you might get numbers that make sense. 200MB for a HDD does not; bogus scores.

The tests you've done are tests on the DRAM of the harddrive and the SATA interface; not the mechanical harddrive itself.
 
sub.mesa said:
Try increasing the test size to 1000MB+ and you might get numbers that make sense. 200MB for a HDD does not; bogus scores.

The tests you've done are tests on the DRAM of the harddrive and the SATA interface; not the mechanical harddrive itself.
Click to expand...

added a 4000mb run just for you. :D
 
Those look much better! Though strange that the RAID0 read scores are way lower than the writes. Could be an affect of the 'write cache' option that buffers the data through your RAM. To check this, set the test size even higher (8000MB) and see if the sequential write score changes; if it becomes a tad lower then likely the effects of the RAM buffercache artificially increased sequential write scores on your RAID0.

Either way, the other two scores of single disks look fine to me. While 6Gbps does not add any real performance, if you would use expanders they would run at higher bandwidth instead compared to 3Gbps disks; the expander would run at 3Gbps as well. Otherwise the difference should be minimal.
 
So lets compare 6GBs in RAID5 to these results using the same benchmark.

It seems my READ speeds are high, but WRITE is pathetic, I wonder how this could be?

My sig is below, but its (3) WD black 640GB 6GB/s drives in raid 5 on an AMD® SB850 Southbridge

CrystalMark4000MB.jpg


CrystalMark1000MB.jpg


CrystalMark50MB.jpg
 
Last edited:
chriss95ta said:
So lets compare 6GBs in RAID5 to these results using the same benchmark.

It seems my READ speeds are high, but WRITE is pathetic, I wonder how this could be?

My sig is below, but its (3) WD black 640GB 6GB/s drives in raid 5 on an AMD® SB850 Southbridge

CrystalMark4000MB.jpg


CrystalMark1000MB.jpg


CrystalMark50MB.jpg
Click to expand...

Pretty much a known quantity at this point that AMD onbaord RAID < Intel onboard RAID and both are significantly worse than quality hardware raid with BBWC and a dedicated parity processor.

Also, I am not exactly sure why it is surprising that hard drives that peak at 150MB/s (let's be fairly generous) when transferring from rotating media (excluding DRAM cache) do not see much of a difference moving from an interface that gives 285-300MB/s real world transfer rates to a newer interface giving 550-600MB/s. 6gbps SATA on traditional drives is pretty much a way to make a few margin dollars off of people who do not know better.
 
pjkenned said:
Pretty much a known quantity at this point that AMD onbaord RAID < Intel onboard RAID and both are significantly worse than quality hardware raid with BBWC and a dedicated parity processor.

Also, I am not exactly sure why it is surprising that hard drives that peak at 150MB/s (let's be fairly generous) when transferring from rotating media (excluding DRAM cache) do not see much of a difference moving from an interface that gives 285-300MB/s real world transfer rates to a newer interface giving 550-600MB/s. 6gbps SATA on traditional drives is pretty much a way to make a few margin dollars off of people who do not know better.
Click to expand...

SO bsically you are saying the new 6GB/s WD drives with 64MB cache I have at most would peak at 150MB/s???

And my reads are significantly higher (192MB/s) than the originally posted drives in the 4000MB bench mark, that is pretty damn fast I take it?>

And my reads at 1000MB are just a bit lower than the post 128GB SSD drive. You have me confused a bit here, my reads are very fast from what I see, writes are just low.
 
#1 my understanding is WD revised the 640GB blacks with a lower density 3 x 220GB platter design, replacing the previous 2 x 320GB platter design. that dropped sequential I/O. you may have the newer rev.

#2 motherboard based RAID5 isn't exactly an objective test if you're trying to determine peak rates. you should start by benching each drive individually, and then add them to a RAID0 array to see how the controller scales. also try a test with the drives set as JBOD and use Windows software RAID0, bypassing the controller's RAID functionality.

#3 your 50MB CDM test only tested the cache

#4 your RAID5 writes are slow because its motherboard raid
 
SO bsically you are saying the new 6GB/s WD drives with 64MB cache I have at most would peak at 150MB/s???
Click to expand...
no that part wasnt specifically directed at you, he was remarking in general about hard drives not seeing any difference between SATA 3 and SATA 2 interfaces because that is not where the bottleneck lies. the average physical hard drive peaks between 100-200mb/sec read speed. SATA2 has enough bandwidth to support just under 300mb/sec. it doesnt matter if you were using an interface with 10gb/sec bandwidth if the drive itself can only read at 150mb/sec. this concept shouldnt really be news to anyone.

you also cant really compare disks in RAID0 to disks in RAID5. you can only compare the RAID levels themselves, but thats not what this thread was about...
 
ghost6303 said:
no that part wasnt specifically directed at you, he was remarking in general about hard drives not seeing any difference between SATA 3 and SATA 2 interfaces because that is not where the bottleneck lies. the average physical hard drive peaks between 100-200mb/sec read speed. SATA2 has enough bandwidth to support just under 300mb/sec. it doesnt matter if you were using an interface with 10gb/sec bandwidth if the drive itself can only read at 150mb/sec. this concept shouldnt really be news to anyone.

you also cant really compare disks in RAID0 to disks in RAID5. you can only compare the RAID levels themselves, but thats not what this thread was about...
Click to expand...

Thanks for the info, my system seems to be very fast and is redundant which is what I wanted.

Sorry to cause the confusion.
 
heres the deal guys.
there is no native SATA 6GB/s onboard controller for intel. the things people are using are a hacked on solution for marketing purposes. It is a disgrace. the chipset does not have the bandwidth necessary to do the job. they are using "spare" DMI and PCI-E lanes to emulate the bus. doesnt work worth a shit. they are giving the spec a bad name with these terrible implementations.
the AMD version sucks. period.

when SB comes, with native 6gb/s spec, you shall see the gains.
 
You must log in or register to reply here.
Back
Top