1 2 Previous Next 17 Replies Latest reply on Feb 4, 2011 1:36 PM by parsec

    Firmware Update 2CV102M3 Results

    retiredfields

      I have 2 X-25M 80gb in Intel RAID 0. Here are before and after results for the 102M3 firmware update:

       

      Before:

       

      before.jpg

       

      After:

      after.jpg

       

      No significant difference, as far as I can tell.

       

      I'm running Win7 64 and this comment from the update readme file did not happen:

       

      Note: For computers booting into Windows 7 after a firmware update,
      a message appears when the operating system starts that prompts
      you to restart the computer. Restart when prompted.



       

      FYI

       

      Ken

        • 1. Re: Firmware Update 2CV102M3 Results
          mistermokkori

          retiredfields wrote:

           

          I'm running Win7 64 and this comment from the update readme file did not happen:

           

          Note: For computers booting into Windows 7 after a firmware update,
          a message appears when the operating system starts that prompts
          you to restart the computer. Restart when prompted.

           

          that message is from windows redetecting the ssd due to the firmware update.  but since you're running the drives in an array, windows won't see that they've changed.

          • 2. Re: Firmware Update 2CV102M3 Results
            retiredfields

            Good point.....Thanks

            • 3. 2R0 160Gb X-25M
              Mitchb

              I saw virtually no difference in performance:

               

              intel_r20.jpg

               

              MSI main board

              980X 4.3GHz

              12Gb Corsair Dominator 2000

              • 4. Re: 2R0 160Gb X-25M
                parsec

                Well, it's not a given that this firmware update will increase performance as measured by 'Crystal or other benchmarking tools.  For example, the update could be fixes or enhancements to wear-leveling, responses to TRIM commands, or bug fixes in the controller code that most of us would not understand without some education.

                 

                This is what Intel included in the Release Notes for the 2CV102M3 Firmware update:

                 

                This firmware revision fixes enumeration and slow-boot issues on SATA 6Gb/s controllers, adds improvements to S.M.A.R.T. attributes for more accurate reporting of drive health, improves NCQ capability, and fixes possible drive hangs when reading S.M.A.R.T. self-test log.

                 

                Nothing that is directlyly performance related, except for NCQ capability, which only comes into play with SSDs when the latency of other parts of the PC causes the SSD to wait for them.  Perhaps they aren't mentioning everything, but they didn't mention enhancing the "user experience".  Oh well...

                • 5. Re: Firmware Update 2CV102M3 Results
                  parsec

                  Ok, first time I've done a firmware update on any of my Intel SSDs.  Worked fine, easy, worked on my PC in AHCI mode.  It was a newer, unused 80GB G2, going from 02HD to 02M3.  I put a few files (jpg) on it just to see if they would be cleared or damaged, they were preserved and just fine.  Here are before and after benchmarks:

                   

                  as-ssd INTEL SSDG2GC MBs i7 10.26.2010.png

                   

                  as-ssd  INTEL SSDG2GC MBs 1.31.2011.png

                   

                  Although the differences are small, and my SSD is an 80GB vs redux' 160 GB, I'll compare them anyway.

                   

                  I lost a little in both sequential read and write speeds, while redux gained a bit in each.

                   

                  Redux lost a little more than I did in 4k read and write, and while mine both diminished, my read speed was barely touch.

                   

                  In 4k with 64 threads, we both increased a decent amount in read speed, and just a touch increase in write.

                   

                  Our read and write access times are the opposite of each other, redux writes are faster than the reads, mine vice versa.  My new read access time is pretty good, even if I lost 0.003ms.

                   

                  If we can accept these results at face value, I traded some sequential read and write performance, and some 4k write performance for a relatively large gain in 4k 64 thread (NCQ depth) read speed.  Given that NCQ was developed for HDDs to reduce head movement and latency due to their functional realities, while those issues are non-existent in SSDs, I must ask what is the point in optimizing this in SSDs?

                   

                  I've read that SSDs use NCQ to keep busy while they wait for other systems of a PC that are to busy to accept data the SSD is ready to send to them.  So with modern CPUs and DRAM memory, this still occurs?  I really don't know, but that would surprise me if true.  I'm just questioning and trying to learn.

                  • 6. Re: Firmware Update 2CV102M3 Results
                    redux

                    NCQ in an Intel SSD enables concurrency in the drive so that up to 32 commands can be executed in parallel.

                     

                    One of the revisions listed for with the f/w is improved NCQ capability, which is why 4K/64 performance has improved.

                    • 7. Re: Firmware Update 2CV102M3 Results
                      parsec

                      Retiredfields, those are some amazing numbers, most of them nearly exactly twice that of a single 80GB G2 Intel SSD.  I see you are using iaStor, so I imagine your mother board has an ICH10R chip.  Do you do anything to your SSD to compensate for the lack of TRIM due to RAID?  Do you have an OS on your RAID setup?  If so, was it compromised by the update, or was it untouched?

                       

                      Although you aren't seeing increases (actually, slight decreases) in some of the standard SSD performance specs, like sequential read and write speeds, Intel's SSD performance philosophy is to tune for real world performance enhancement.  The ability to perform multiple simultaneous I/O requests via NCQ at a greater rate then before with this update is more tuning for enhanced real world use.  Although it is more difficult to relate to or appreciate increased speed with multiply queued commands vs sequential speeds, the former will actually be of more benefit to a user, IMO.  We need to train ourselves to see beyond the simple "drag race" numbers and appreciate the complexities of these devices.

                      • 8. Re: Firmware Update 2CV102M3 Results
                        parsec

                        Yes redux, I wasn't seeing it in that way.  I'm aware that the NCQ enhancement is the major performance upgrade in this update.  Not that this will impress the sequential R/W speed demons, but I don't care.  Seeing how those specs dominate most SSD marketing, to sell to the less informed consumer, I'm pleased to see Intel is not going in that direction.  OTOH, the specs of the new 510 SSD do seem to be going in that direction.  I can understand doing that, but it's a shame they feel they need to do so, if that is the case.

                        • 9. Re: Firmware Update 2CV102M3 Results
                          CarrotThree

                          Before

                           

                          Intel postville raid cdm.png

                           

                          After ,NCQ is looking better

                          cdm 290.PNG

                          • 10. Re: Firmware Update 2CV102M3 Results
                            retiredfields

                            Yes, ICH10 on an EVGA X58 motherboard.

                             

                            I posted this http://communities.intel.com/thread/14568 about my work around for no TRIM. Still seems to be working fine. I also leave about 1/3 of the array unallocated to give the drive firmware plenty of room to perform its cleanup.

                             

                            http://communities.intel.com/thread/14568

                             

                            I am using Win7 64 and there was no problem doing the firmware update. Changed BIOS to IDE legacy mode, did the update on each drive, and then changed back to RAID mode. I did have a current image backup, just in case......

                            • 11. Re: Firmware Update 2CV102M3 Results
                              parsec

                              Retiredfields, it sure did work, with your RAID 0 write speed being exactly twice the typical 78 - 80 MB/s single 80GB G2 SSD write speed, after your free space consolidation.  Of course, we could ask if the cost of the rewriting using Perfect Disk 11 is more than that if TRIM was available, or if the Toolbox Optimizer could have been used.  Even if it is overall, I'd say it's a moot point given the longevity of the NAND chips that Intel uses.  You could check the SMART attributes and see what your Host Writes and wear indicator are showing, but I doubt it makes any real difference.

                               

                              I'm not a RAID user so I don't know if NCQ is functioning on 'drives or volumes in a RAID array.  It seems that it must be, although your 4k - 64 Thread read spec is less than twice my single 80GB G2, but not by that much, and the same with the 4k - 64 writes.  It's interesting that your write access time is less than the read access time, which was the case with another forum members 160GB G2 SSD, while my single 80GB G2 has faster read than write access speed.

                               

                              I also wonder how the benchmarking tools can measure speeds that are greater than the SATA 2 interface speed, generally specified as 300MB/s maximum.  I'm not saying they are false or inaccurate, I don't know either way, I'm just curious.  Wait, what am I saying, it must be that the tool "sees" the RAID 0 volume physically connected via two cables as one 'drive/volume.  Is that the case?

                               

                              I was wondering if the data on the SSD after the firmware update was always corrupted, as Intel recommends (seems more like a warning) having an image back-up done before the FW update.  Now wouldn't it be nice if there was a stand-alone version of the Intel SSD Toolbox Optimizer program, that was part of an .iso file like the FW update utility, that could be run on RAID arrayed SSDs that were set to IDE mode for the optimization, just like the FW update.  Your Perfect Disk method seems to make that unnecessary.

                              • 12. Re: Firmware Update 2CV102M3 Results

                                Just wanted to post that on none of the systems I manage did ACHI work.  I had to set to IDE mode to update the firmware.  We use primarily HP Elite Desktop SFF 8100's, HP Business laptop 2540p's(laptop).

                                • 13. Re: Firmware Update 2CV102M3 Results
                                  parsec

                                  In my case, I have an ASUS SaberTooth X58 mother board, using the Intel RST iaStor driver, connected to the Intel ICH10R SATA 2 ports, with Windows 7 64 bit.  I decided to experiment with the update, and left it in AHCI mode.

                                   

                                  Your experience, mjm, is not unusual, it seems most users must change to IDE or legacy/compatibility mode.

                                   

                                  Apparently, my overall configuration was just right to allow the update to occur in AHCI mode.

                                  • 14. Re: Firmware Update 2CV102M3 Results
                                    parsec

                                    CarrotThree, do you have a RAID setup of SSD's, or what?  A bit unusual that you did not loose any sequential read speed after the update, and actually increased!  We do see the apparently common loss of 4k write speeds, and the great gain in 4k NCQ read speeds.  Are you pleased with the overall results?

                                    1 2 Previous Next