What do your hard disks score on xbench?

[nightfly13]

Depressingly, the stock 250gb drive that shipped with my first gen 2.66 quad system (dual dual) comes up with a Disk Test of 33.41. Ick! This is my system drive! I've been itching to get a SSD in here because I've felt my hard drive speeds are lackluster - and now I see how appalling they really are! I also have some 500gb Samsung Spinpoints that only score 37-40 and the newest drive my modest collection, a year old 640gb Spinpoint scores best around 90.

I'm sure a SSD will help feel and performance (planning for the 80gb Intel) but is there anything I can do about my archival data drives?

Also does anyone know offhand how easy it is to utilize the SATA ports on the logicboard of the 1,1 Mac Pros? Any link would be appreciated - so I don't have to yank the 250gb out to install the SDD but just slam it in there somewhere.

 

[hyram]

Ok... here's the results for 8 x 1 TB WD Blacks in RAID 5 with a Highpoint 4322 controller:

Results 571.59
System Info
Xbench Version 1.3
System Version 10.4.11 (8S2167)
Physical RAM 8192 MB
Model MacPro1,1
Drive Type HPT DISK 0_0
Disk Test 571.59
Sequential 495.66
Uncached Write 532.11 326.71 MB/sec [4K blocks]
Uncached Write 1153.58 652.69 MB/sec [256K blocks]
Uncached Read 211.75 61.97 MB/sec [4K blocks]
Uncached Read 1663.13 835.88 MB/sec [256K blocks]
Random 675.01
Uncached Write 370.79 39.25 MB/sec [4K blocks]
Uncached Write 1086.92 347.96 MB/sec [256K blocks]
Uncached Read 1930.86 13.68 MB/sec [4K blocks]
Uncached Read 558.35 103.61 MB/sec [256K blocks]

Not as good as the 6 drives in RAID 0, but a whole lot more reliable.


Here's the results of Disktester area scan:

---------------- Averages for "RAID" (512MB/32MB, 10 iterations) -------------
Area (6.36TB) Write MB/sec Read MB/sec
0% 751 480
5% 758 474
10% 743 480
15% 747 480
20% 728 458
25% 733 467
30% 724 452
35% 703 449
40% 714 446
45% 705 449
50% 684 434
55% 692 423
60% 686 414
65% 648 404
70% 635 389
75% 622 376
80% 584 355
85% 573 352
90% 523 320
95% 501 304
100% 467 278
Average write speed across the volume: 663MB/sec
Average read speed across the volume: 414MB/sec

And below is the Black Magic Speed Test.

I think I can live with these numbers.

hyram

 

[Tesselator]

Ok... here's the results for 8 x 1 TB WD Blacks in RAID 5 with a Highpoint 4322 controller:

Here's the results of Disktester area scan:

---------------- Averages for "RAID" (512MB/32MB, 10 iterations) -------------
Area (6.36TB)    Write MB/sec    Read MB/sec
       0%             751            480    
       5%             758            474    
      10%             743            480    
      15%             747            480    
      20%             728            458    
      25%             733            467    
      30%             724            452    
      35%             703            449    
      40%             714            446    
      45%             705            449    
      50%             684            434    
      55%             692            423    
      60%             686            414    
      65%             648            404    
      70%             635            389    
      75%             622            376    
      80%             584            355    
      85%             573            352    
      90%             523            320    
      95%             501            304    
     100%             467            278    
Average write speed across the volume: 663MB/sec
Average read speed across the volume: 414MB/sec

Really? I remember my Level 5's being faster than this. This is not significantly different than my current RAID0 3-drive configuration. Hmm...

 

[nanofrog]

hyram, is the cache still switched off?

 

[hyram]

The cache is set to write back. Sector size is 512B, block size is 256k, NCQ is enabled INT 13 support is enabled (not sure about this one). These are the same settings I had when set as RAID 0.

I'm open to suggestions to try and improve this.

hyram

 

[nanofrog]

The cache is set to write back. Sector size is 512B, block size is 256k, NCQ is enabled INT 13 support is enabled (not sure about this one). These are the same settings I had when set as RAID 0.

I'm open to suggestions to try and improve this.

hyram

SAS uses TCQ, not NCQ.

So set NCQ = Disabled. You'll should see a difference.
(BTW, don't try to enable TCQ either).

 

[Virtuoso]

Caldigit RAID5 with 4x WD RE2 drives vs Corsair SSD 256GB...

 

[hyram]

SAS uses TCQ, not NCQ.

So set NCQ = Disabled. You'll should see a difference.
(BTW, don't try to enable TCQ either).

Thanks for the suggestion.. I'll give it a try as soon as I can. And as far as I know I can't enable TCQ.

hyram

 

[aibo]

If anyone wasn't already convinced SSD is the way to go, this thread should do it. People are needing 6+ drive RAIDs just to reach the performance level of a single Intel/OCZ SSD.

 

[Tesselator]

If anyone wasn't already convinced SSD is the way to go, this thread should do it. People are needing 6+ drive RAIDs just to reach the performance level of a single Intel/OCZ SSD.

• 2-SSD RAID0 Test Score: ---------------- 438.45
• Single SSD Disk Test Result: ------------ 267.07
• Single SSD Disk Test Result: ------------ 249.52
• 8-Drive RAID0 Disk Test Result: -------- 782.31
• 6-Drive RAID0 Disk Test Results: ------- 617.08
• 4-Drive WD 640GB RAID Test Results: -- 238
• 4-Drive WD VR RAID Test Results: ------ 439.75
• 3-Drive GREEN RAID Test Results: ------ 176.78
• 2-Drive BLACK RAID Test Results: ------ 168.69

Q. Items B and C have similar results to what other items in the list?

Q. Which items in the list have better test results than either B or C?

 

[hyram]

"If anyone wasn't already convinced SSD is the way to go, this thread should do it. People are needing 6+ drive RAIDs just to reach the performance level of a single Intel/OCZ SSD."

Debatable… and only if speed is you're only criteria. I'm working with HD video, I have to trade off speed AND capacity. When file sizes approach 200-300 GB a moderate sized RAID fills up fast, a SSD is already full. And the $/GB is not even close… how much would I have to spend to get an SSD in the neighborhood of 7TB???

Don’t get me wrong… SSD has it’s place; I’m toying with the idea of an SSD for a boot/app drive. But I’ll probable wait for pricess to drop and capacity to increase a little more first.

But this RAID is for data only.

hyram

 

[VirtualRain]

I'm not sure Xbench adequately captures the impact that sub-millisecond access times have on every-day use.

I suspect the 4K Random read performance benchmark is the best measure of the impact that SSD's can have in most every-day applications... in some cases, an order of magnitude improvement. That's the only thing I can think of to make me such a fan-boy (besides the cool-aid that Intel keeps sending me! )

 

[Tesselator]

Hehe... well at least you're straight about it. That's kewl!

Do keep in mind tho that most of the accesses in a 3 or 4 drive RAID will be sub-millisecond too. Mine are.

 

[hyram]

SAS uses TCQ, not NCQ.

So set NCQ = Disabled. You'll should see a difference.

It made absolutely no difference it NCQ is set or not.

I've been playing around with the test... various chunk sizes and file sizes, and I find this has the biggest impact. For example: iIf I set it to what was used in the AMUG review 128MB/10GB then I get almost identical performance to what AMUG reports. Difference is they are running Samsung's while I have the WD drives. Bt not a very big difference in the results. I'm going to try and do a more thurough study on this but it will have to wait a few days.

Tesselator, when I look at your results, they show fairly flat accross the whole drive surface. Most tests I've seen usually show some reduction in speed as the drive fills up. How are you results possible???

hyram

 

[nanofrog]

It made absolutely no difference it NCQ is set or not.

I've been playing around with the test... various chunk sizes and file sizes, and I find this has the biggest impact. For example: iIf I set it to what was used in the AMUG review 128MB/10GB then I get almost identical performance to what AMUG reports. Difference is they are running Samsung's while I have the WD drives. Bt not a very big difference in the results. I'm going to try and do a more thurough study on this but it will have to wait a few days.

It can on SAS cards. At least enough you can notice, and is supported by tests. It's not too much now on the Areca's I've used (since firmware 1.45). But as your's is a little different (HighPoint), I figured it was worth a shot.

Stripe size can make a big difference, but had assumed you'd already settled on that (dialed it in for your usage/gotten advice elsewhere on it), since you hadn't asked. Oops.

 

[Tesselator]

Tesselator, when I look at your results, they show fairly flat accross the whole drive surface. Most tests I've seen usually show some reduction in speed as the drive fills up. How are you results possible???

hyram

I have no idea. It's just the way those particular drives profile. Keep in mind that these are GREEN drives and from what I read once the spindle speed is not only low 5900 RPM (not 5400 like some resale sites list it as) according to one Japanese source, it's variable speed - which sounds strange to me. So I wonder if those things have anything to do with it? Which benchmarks are you referring to specifically?

But if you look here: https://forums.macrumors.com/posts/7673878/ you will see the tapering off that you're talking about everywhere the drives aren't slamming up against some other limit.

 

[nanofrog]

Keep in mind that these are GREEN drives and from what I read once the spindle speed is not only low 5900 RPM (not 5400 like some resale sites list it as) according to one Japanese source, it's variable speed - which sounds strange to me.

This is what I'd expect for that drive. WD is doing the same thing to boost performance on their "Green" lines.

From what I've seen, they typically spin between 5400 and 7200rpm (getting faster as it gets to the inner tracks).

 

[hyram]

Mostly from AMUG reviews, but I've seen it on other sites as well. I uess I hadn't scrolled down enough on your graphs because the larger file sizes in your data do exhibit this as well.

I'm not sure I have much here to complain about, most of the time the rates at full are a little less than half the rate of empty drives. For mine the rates are a little more than half at full.

Write (MB/sec) Read (MB/sec)
706 689
695 678
677 659
668 647
643 624
621 603
585 564
544 522
501 487
442 426
370 359


This is with 10GB files in 128MB chunks.

It'll take me while,but I plan to do more thorough testing of the file/chunk size.

Mr. Frog... it didn't hurt to try it. I've just left it off as I can't see a difference with or without.

Thanks,

hyram

 

[nanofrog]

Mr. Frog... it didn't hurt to try it. I've just left it off as I can't see a difference with or without.

Thanks,

hyram

Mr.? Where?

Take it as a good thing. That means the firmware has gotten good enough that the difference has diminished, though not totally gone, as the processor is still SAS optimized. It took a while, but Areca seems to have gotten it sorted rather well with v1.45 (now at v1.46, but I've not run SATA on this revision). So perhaps the difference has dropped even further.

I wouldn't worry about it at any rate.

 

[Tesselator]

I'm not sure I have much here to complain about, most of the time the rates at full are a little less than half the rate of empty drives. For mine the rates are a little more than half at full.

Write (MB/sec)	Read (MB/sec)
706	             689
695	             678
677	             659
668	             647
643	             624
621	             603
585	             564
544	             522
501	             487
442	             426
370	             359

This is with 10GB files in 128MB chunks.

It'll take me while,but I plan to do more thorough testing of the file/chunk size.

Yeah, this is true. As a drive or raid set approaches 80% full performance begins to suffer a lot! I guess it's the combination of volume & file fragmentation along with likely having to work more with it's inner slower sectors.

The remedy these days is pretty straight forward. Use a big-ass RAID set and don't let it get more than 50% full. That way you're always on the cusp of it's high performance cylinder bands.

 

[Dr.Pants]

Coming in with a Hitachi Deskstar at 500GB capacity and 7200RPM, itssss...

DR.PANTS!!!

Disk Test 71.01
Sequential 108.33
Uncached Write 105.68 64.89 MB/sec [4K blocks]
Uncached Write 103.21 58.39 MB/sec [256K blocks]
Uncached Read 108.81 31.84 MB/sec [4K blocks]
Uncached Read 116.52 58.56 MB/sec [256K blocks]
Random 52.82
Uncached Write 20.32 2.15 MB/sec [4K blocks]
Uncached Write 111.03 35.55 MB/sec [256K blocks]
Uncached Read 95.88 0.68 MB/sec [4K blocks]
Uncached Read 141.37 26.23 MB/sec [256K blocks]

Anything look "off" on this to the initiated

 

[Virtuoso]

WD is doing the same thing to boost performance on their "Green" lines. From what I've seen, they typically spin between 5400 and 7200rpm (getting faster as it gets to the inner tracks).

I think this has been exposed as Marketing 'misdirection'. The Green drives are a fixed speed, not a variable speed, of either 5400 and 7200, although apparently all the ones currently in the field run at 5400. From the SilentPC review:-

Most of a drive's power is consumed by the motor that spins the disk inside the drive. Reduce the speed of the disk, and you reduce the amount of power required. However, Western Digital doesn't want to say that they're selling 5,400 RPM drives — those became second class in the desktop market years ago. Instead, they rate the drive's speed as "IntelliPower" and take pains to emphasize that there are other factors that affect performance.

Western Digital has caught a lot of flak for withholding the rotation speed of the Green Power, especially when the product was first launched and the marketing material listed the rotation speed as 5,400-7,200 RPM. This led some to speculate that the rotation speed changed dynamically during use — which would have been an impressive engineering feat had it been true. The reality is revealed by a sentence that Western Digital added to the description of IntelliPower: "For each GreenPower™ drive model, WD may use a different, invariable RPM." In other words, Western Digital reserves the right to release both 5,400 RPM and 7,200 RPM drives under the Green Power name — without telling you which are which.

We were able to confirm that our 750 GB Green Power had a spindle speed of 5,400 RPM by doing frequency analysis on a sound recording of it. Why sound? Sound is vibration; the pitch of the sound corresponds to the frequency of the vibration. Hard drives vibrate at the speed of their motor, so they produce a noise at the same frequency as their rotation speed. Our sample had a sharp spike at exactly 90 Hz (cycles per second). Multiplying that number by 60 (to get cycles per minute) yielded a measured rotation speed of 5,400 RPM.

It's possible that other Green Power models use a higher spindle speed — but we doubt it. Storage Review tested the 1 TB version of the drive and determined that that model also spun at 5,400 RPM based on a calculation of the drive's latency compared to a previous Western Digital model. That leaves the 500 GB model — which Western Digital says is even lower power than the larger capacity versions. With the majority of the Green Power's efficiency advantage coming from its lower speed, it seems impossible for the 500 GB model to use a higher rotation speed. It's possible Western Digital intends to release a 7,200 RPM version at some point in the future.

 

[nanofrog]

I think this has been exposed as Marketing 'misdirection'. The Green drives are a fixed speed, not a variable speed, of either 5400 and 7200, although apparently all the ones currently in the field run at 5400. From the SilentPC review:-

Nice catch.

I hadn't spotted this, but I guess I shouldn't be surprised either. Given how drive makers determine capacity, and other values in the specifications.

 

[Tesselator]

I don't think this changes anything. I think this is still just another person who doesn't actually know's opinion. I think it should be weighed as such too. It's for sure not even empirically definitive. For example he has to test the speed by recording vibrations. Yeah, but what was the drive doing at the time and did he try it while the drive was forced to access the inner most cylinders? The only thing in all this text that may not be wholly speculative is the mention of WD adding the "For each GreenPower™ drive model, WD may use a different, invariable RPM." text to their site - which I haven't seen yet and don't know the context.

So for me it could still likely be either way and no one knows what the spindle speed of the Samsung F2 EcoGreen HD154UI is yet.

 

[nanofrog]

I don't think this changes anything. I think this is still just another person who doesn't actually know's opinion. I think it should be weighed as such too. It's for sure not even empirically definitive. For example he has to test the speed by recording vibrations. Yeah, but what was the drive doing at the time and did he try it while the drive was forced to access the inner most cylinders? The only thing in all this text that may not be wholly speculative is the mention of WD adding the "For each GreenPower™ drive model, WD may use a different, invariable RPM." text to their site - which I haven't seen yet and don't know the context.

So for me it could still likely be either way and no one knows what the spindle speed of the Samsung F2 EcoGreen HD154UI is yet.

Quite possible. I'll recheck the Green data sheets on WD's site, and look for the proviso at the bottom.