Overclock.net › Forums › Components › Hard Drives & Storage › SSD › [Official] Crucial SSD Owners Club
New Posts  All Forums:Forum Nav:

[Official] Crucial SSD Owners Club - Page 153

post #1521 of 1523
Quote:
Originally Posted by error-id10t View Post

My 4k read comment applies to ALL SSDs, they've increased speeds everywhere except there. True, some are slightly faster but they ALL still suck at 4k reads. One of the review sites mentioned they saw improvements in speeds after using it after awhile so keep an eye on that and of course Crucial may be bring a firmware out that speeds it up (most usually do).

I basically agree about the 4K read speeds not increasing much, but there are exceptions. AS SSD is not the only benchmark that matters, it you check SSD reviews at Anandtech, that use IOmeter among others, 4K reads have improved. It's curious that the best AS SSD 4K read score, until recently, was given by the Crucial C300 SSD, in the low +30MB/s range. That was not surpassed until the Samsung 840 Pro hit 37-38MB/s in AS SSD 4K read speed.

The C300 could reach over 30MB/s 4K read speeds in AS SSD, with ONFI 1.0 NAND (100-133MB/s NAND speed), with one of the first SATA III SSD controllers. If newer SSDs with faster NAND and better controllers cannot surpass or even equal it, that implies the firmware was coded to trade 4K read speed for something else, IMO. Is SSD firmware biased for high sequential (marketing) speeds?
post #1522 of 1523
Quote:
Originally Posted by parsec View Post

An interesting AS SSD result for the M500's in RAID 0. A higher overall score than I've ever seen for m4s in RAID 0.

I'm sure you know but that's mainly due to the 4k-64thrd increase which AS-SSD for some reason loves over anything else (if they'd drop that to 8, 16 or 32 then it'd make more sense IMO). Though like you said, AS-SSD is just 1 bench and reviews have shown the drive to match the fastest ones out there (bar 1 or 2 reviews which I question) and destroys M4 basically.

Agree about the latency, liking that.
Quote:
Originally Posted by parsec View Post

Anyone using the new IRST 12 drivers needs to know Intel has added a few new performance tuning features, that are available by running the IRST 12 GUI in Windows:



The default setting is Read only. I have not tested each setting to see how they differ.

You want Write Back otherwise 4k writes are throttled. The rest of the options give around the same speeds (for me at least).
Quote:
Originally Posted by parsec View Post

Is SSD firmware biased for high sequential (marketing) speeds?

This would be my guess, big numbers look prettier.
Mine
(20 items)
 
  
CPUMotherboardGraphicsGraphics
2600k ASUS Z77-V EVGA 670 4GB SC+ EVGA 670 4GB SC+ 
RAMHard DriveOptical DriveCooling
G.Skill Ares F3-2133C9D-8GAB 2x Crucial M4 256GB (RAID0) Lite-On iHBS212 RX360 
CoolingCoolingCoolingCooling
Raystorm block EK Full Cover VGA Block EK-FC670 Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump 
OSMonitorKeyboardPower
Win8 Pro x64 BenQ XL2410T 120hz MX5500 Seasonic X-760 
CaseAudio
Corsair 800D Creative Sound Blaster X-Fi Titanium HD 
  hide details  
Reply
Mine
(20 items)
 
  
CPUMotherboardGraphicsGraphics
2600k ASUS Z77-V EVGA 670 4GB SC+ EVGA 670 4GB SC+ 
RAMHard DriveOptical DriveCooling
G.Skill Ares F3-2133C9D-8GAB 2x Crucial M4 256GB (RAID0) Lite-On iHBS212 RX360 
CoolingCoolingCoolingCooling
Raystorm block EK Full Cover VGA Block EK-FC670 Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump 
OSMonitorKeyboardPower
Win8 Pro x64 BenQ XL2410T 120hz MX5500 Seasonic X-760 
CaseAudio
Corsair 800D Creative Sound Blaster X-Fi Titanium HD 
  hide details  
Reply
post #1523 of 1523
error-id10t, you are right, I agree. That slipped my mind as I was thinking about IRST 12. I read in a forum, that had the AS SSD programmer posting in it, that he felt the 4K-64Thrd test is a very difficult torture test for SSDs.

That does not seem to be the case, as IMO SSD firmware has been biased to provide high speeds in that test, to jack up the overall score, since otherwise the new products would not score much better. When one manufacture does that, the others must follow, or look bad. First it was sequential read that was increased (which the C300 lacked), then sequential write, and now high queue depth speed, at least with enthusiasts.

I agree that testing a consumer SSD at a queue of 64, twice that of the maximum queue of AHCI (32), is a worthless spec in real world PC use. I highly doubt PC users ever exceed queue depths of single digits in actual use. Which is why the issue some m4s have with that test in AS SSD does not affect real world performance. I understand why people are unhappy about it, I would be too, but it's all psychological. I am able to ignore the lousy AS SSD results my Intel 520's in RAID 0 provide, but that's because in use they are equal or better than SSDs scoring almost twice what they do, which I also use.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: SSD
Overclock.net › Forums › Components › Hard Drives & Storage › SSD › [Official] Crucial SSD Owners Club