Nforce 'SataIDE'-Driver Benchmarking [Updated: 19-05-2008]

[TwL]

Hello, everyone.

Well, since been all the time having conversation about Sata drivers in Nforce systems with Fernando I though how about I make them completly own post here to have full featured compare to Windows Vista & XP systems Sata speeds. I will continue this topic all the time as new drivers popup I will hit them to same systems as below posts & Benchmark them in list. Also will be adding HD Tach v3.0.4.0 & HD Tune testing.


Command Information & what to believe...
Well, I assume you do you own decesions of which driver to use. Generally all drivers jump in between value A & B and this makes a lot of difference when we are comparing an driver which is unstable or stable like difference between v9.92 to v9.98 on vista for example 9.92 is stable after 30 seconds while in v9.98 same result is instant making no difference in games or anything big on load having more features than v9.98. the values on Everest benchmarks are like 5-10 seconds from began to read while. *Update* there is also now stability sections to cover the idea how the driver functions.

Either way, these are just directions how all drivers are acting with this specific benchmarking.


About NcQ & Advanced Caching-Benchmark:
Any Benchmarks shown here are done with NcQ (Command Queueing) Disabled because there is no REAL measure of this speed up system. System is designed for multible loads sequences and/or files loading speed-up and in any benchmark I have this cannot be measured except seeing false speed drop on drivers. However, this feature will always increase your system speed on load & always should be enabled when possible. This affects in everything even while benchmarking results with NcQ are lower. This feature simply wasn't designed to be used as an single file Copy/move/read system.

Advanced HDD caching on SATA/SATA-II was turned on, on every benchmark no matter was it turned off by drivers, unless incapable of doing so like situation benchmarking 'Standard IDE Controller'.


Hardware & Operating systems
Windows XP Service Pack 2 [x64]
Windows XP Service Pack 3 [x86]
Windows Vista - Service Pack 1 [x64]
Processor: AMD Athlon X2 4600+ (2,4Ghz)
Memory: 4GB G.Skill 4-4-3-5-3-16 DDR2 800Mhz
Harddrive: Seagate 320Gb Model: ST3320620AS 3.AA SATA-II (SataIDE)

Previous and below all benchmarks done with this board:

[TwL]

Windows XP x64(64Bit) - Service Pack 2 - Section (Updated: 08-05-2008)

Benchmarking screens:
v6.86 (NIC v65.21): Full Scale 'Everest' *WHQL*
v6.99 (NIC v65.21): Full Scale 'Everest' *WHQL*
v9.99.09 (NIC v65.21): Full Scale 'Everest' *WHQL*
v10.3.0.16 (NIC v65.21): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0
v10.3.0.21 (NIC v67.76.1): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0 - HD Tune v3.00

General Notes:
v6.86.00 -> goes in clean unstabile, but fast & includes S.M.A.R.T. & NcQ Support.
(Works with 'nvata.sys' or 'nvata64.sys' depending on system.)
v6.99.00 -> goes in clean unstabile, but fast & includes S.M.A.R.T. & NcQ Support.
(Works with 'nvata.sys' or 'nvata64.sys' depending on system.)
v9.99.09 -> Goes in Clean doesn't Include S.M.A.R.T. Support. Includes the First 'NEW' Drive detection system to see to HDD Naming layer.
(Works 'nvgts.sys' & 'nvgts64.sys' Integrated to system. Disables 'nvata.sys' or 'nvata64.sys' and This cannot be automatically downgraded to v6.xx of drivers.)
v10.1.0.xx -> These driver has been until v10.1.0.28 very very unstabile (however on RAID systems very good which they were designed to), but now there is v10.1.0.28 out for XP and it will be reviewed because it is perfectly stabile as far testing goes, fast & includes NcQ support (Aka. other side of AHCI as v10.3.0.21 has Hot Plug/swap support but no NcQ). NcQ is very very welcome feature as it manages to queue multible files and commands simultanously and system load, Databases & something like game loading is incredible much faster in this manner even while HDD top speeds would be lower as on use by benchmarks.
(Works 'nvgts.sys' & 'nvgts64.sys' Integrated to system. Disables 'nvata.sys' or 'nvata64.sys' and This cannot be automatically downgraded to v6.xx of drivers.) (Includes NcQ Support, No Hot Plug/Swap supports, Read/Write Caches Enabled by Default, no SMART Supports.)
v10.3.0.16 -> Has an Issue after Drive is detected & crashes Fatal Error 1 times after first reboot after drive detection. Otherwise Stabile. Includes the 'NEW' Drive detection system to see to HDD Naming layer.
(Works 'nvgts.sys' & 'nvgts64.sys' Integrated to system. Disables 'nvata.sys' or 'nvata64.sys' and This cannot be automatically downgraded to v6.xx of drivers.) ([/color]no NcQ Support, Hot Plug/Swap support enabled, Read/Write Caches Enabled by Default, no SMART Supports.)

Note: drivers v6.xx used 'nvrdx64.sys' and driver above v9.92 uses 'nvrd64.sys' (not Windows default 'nvraid.sys' x64 which is left behind still) including mentioned SataIDE driver depending on system for all RAID systems.


Windows XP x86(32Bit) - Service Pack 3 - Section (Updated: 08-05-2008)

Here we can see perfect example how big difference there is at x86 and x64 Architectures still today.
The drivers acts nearly completly like day & night while they are same drivers in all sense. This
testing was good to be done & I think we all learn something about what is the truth behind the scenes
a bit.

Benchmarking screens:
Standart IDE Controller [MS] (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v6.86 (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v6.91 (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v6.99 (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v9.95 (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v9.99.09-[Modded by Fernando] (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v9.99.09-[Modded by Fernando] (NIC v65.76): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v10.1.0.12 (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v10.3.0.16 (NIC v65.21): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v10.3.0.16 (NIC v67.76): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v10.3.0.21 (NIC v67.76.1): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00


General Notes:
Standart IDE Controller [MS]: Gives incredible unstabile base, but feels open & is fast benchmark shows 211.xxMB/s buffered reading speeds, but as seen on the secondary stability tests it is very unstabile however faster than NVIDIA v6.86 indeed.
(Supports S.M.A.R.T. no Native Command Queue(NcQ) or Read cache supports included with this however I think it is caching correctly anyway.) (Function driver file: 'pciide.sys')

v6.86: Very Stabile driver as seen on stability tests. Benchmarks still over the edge around 200MB/s and seems very stabile on all other tests too.
(This driver supports Read caching(Defaults ON), NcQ(Defaults ON) & S.M.A.R.T. support lags some supports on External drive read caching (aka. USB Sata drives).) (Function driver file: 'nvata.sys')

v6.91: Made to only support 5x0 Nforce series. Performs very well still on all boards tested under 430, 590SLI & 630a and seems even while it's not designed to all it still works very fast this driver however is questionable to be used on anything else, but nforce 5x0 boards.
(This driver supports Read caching(Default ON), NcQ(Default ON) & S.M.A.R.T. support. Introduces External SATA USB Drive caching scenarios.) (Function driver file: 'nvata.sys')

v6.99: Last driver designed directly to Windows XP. These are incredible stabile drivers not too fast comparing to earlier ones hits barely 197MB/s to all 3 series of nForce boards (4x0-6x0) and probably the best drivers existing today. Has complete support of all & includes now External buffering to all boards.
(This driver supports Read caching(Default ON or as driver before), NcQ(Default OFF or as driver before) & S.M.A.R.T. support. Introduces External SATA USB Drive caching scenarios to all boards) (Function driver file: 'nvata.sys') (Last driver functioning with 'nvata.sys')

v9.95: Introduces us an new style Hard drive detection system. It reads the Hard drive complete name in this driver & supports also this same function to be used on External HDDs. This also inroduces new Harddrive Write Caching location inside the device driver and not at SATA controller as used to. New driver style at this point isn't too stabile nor it would be fast as seen on benchmark also CPU usage on buffered reading is very high & linear reading is degraded by 5-10MB/s although these are first drivers passing on my system set over 71MB/s & has much lower CPU usage on HD Tune v3.
(No read caching support & NcQ disabled(grey'ed). S.M.A.R.T. Supported.) (Inroduces new style driver: File: 'nvgts.sys' note that previous drivers 'nvata.sys' is still double loading on background and can now be removed by autoruns or by removing the service & then file.)

v9.99.09-[Modded by Fernando]: Well, here is an interesting set from MS actually only tags were to nForce 7x0 board, but thanks to fernando we can now have these on system with correct tags. These still includes S.M.A.R.T. Support for x86 systems and has the v9.95 introduced Drive detection system now although only still with write caching as the v9.95 moved to device. The speed of these are averate but good, averate access has been lowered & stability is very good no more jump jump hehe.. These are probably the best drivers you can add to your system.
(Read caching(Default ON disabled to show from properties), NcQ(Default OFF & #FFFFFE'ed on settings), S.M.A.R.T. Supported & Includes the External new Drive detection from v9.95 style.) (Uses Driver file: 'nvgts.sys')
v9.99.09-[Modded by Fernando]-NIC v67.76: Well, was on expression that on x86 systems this wouldn't have any affect. However when I tested the v10.3.0.16 with the new Network Controller driver it was very very jumpy right after install. So Wanted to test v9.99.09 which I know isn't including rest of the sata drivers optimized to this NIC and you can see clearly now the speed drop. So any unfaithful here can also rest NIC change has incredible affect on hardrive speeds.

v10.1.0.12: Now these are weird ones. These are designed to nForce 630a boards and has spin downs on nForce 4x0,5x0 & generates read errors it seems partitions from chkdsk switch back to older drivers. Includes newest driver set with NcQ supports.
(Read Caching(Defaulted ON Disabled to show on properties), NcQ Supported (Defaults ON Overwrites earlier Driver sets), S.M.A.R.T. Supported, has full device naming & external HD Naming & buffering support) (Uses Driver file: 'nvgts.sys')

v10.3.0.16: Inroduces completly new interface again to Controller. Including unadjustable AAM and 'free fall control' for all other boards than nForce 7x0 series. Also NcQ gets disabled again. This driver is averate on speed, Stabile & should be modded to all boards functions well in 590SLI & 630a tested here too.
(Read Caching(Enabled by default hidden function), NcQ(Disabled & #FFFFFE'ed), S.M.A.R.T. Support(Enabled), New Drive naming deteciton & Including new Sata HDs, Better External HD name detection & External HD Buffering Support) (Uses Driver file: 'nvgts.sys')
v10.3.0.16-NIC v67.72: Note that this driver does not suffer the same lag & degrade performance as in the x64 systems shows clearly the performance degrade & on boot lag. However, you can note clearly how jumpy it becomes by this Network Controller update. It's few times worst on v67.72 and even worst on x64 systems. Also note that v10.3.0.16 was optimized for this NIC check v9.99.06 with this NIC also.

v10.3.0.21-NIC v67.76.1: Driver is insanely stabile & compatible to all. v67.76.1 seems to be functional and good solutions with this. It gives more speed than at Vista with this driver. v67.76.0 how ever is latest recommended still as these two NICs are completly different techs. So please trust that. v10.3.0.16 is still fastest there is for this style of system. v9.99.09 is fastest when considering Single harddrive performance with compined v65.xx NICx. (Read/Write caching at drive properties, No NcQ, S.M.A.R.T. or other AHCI support however hotswap works. This brings new AAM style to use however grey'ed.)

Note: drivers v6.xx used 'nvraid.sys' and driver above v9.92 uses 'nvrd32.sys' (note Windows default 'nvraid.sys' is not used on new drivers which is left behind still) including mentioned SataIDE driver depending on system for all RAID systems.


Windows XP Programs, caching & system set:
Every single benchmark and testing were done in same machine. All XP basic software after installation were on and no other extra software were installed on any state exception of 'Everest' for making Benchmarks. Rest of the drivers were updated so:
-Ethernet Controller: *Update* NIC used on Benchmark Notifyied with the benchmark.
-SMBus: v4.6.3
Other:
-No other devices or drivers were on system or by detected by XP SP2.
-No other software were run in the testing
-Windows XP Caching were(1 Gbyte Filesystem Caching) filled full before any Benchmarks were began to it to not interfere the result & for get the best usage result from system.
-Drivers were installed hardcore way. System was stripped down always to MS Normal IDE drivers after each benchmark & Driver files were deleted before new drivers were installed to make sure there is no left overs at system.

[TwL]

Windows Vista x64(64Bit) - Service Pack 1 - Section (Updated: 16-04-2008)

Benchmarking screens:
Standart IDE Controller [MS] v6.0.6001.18000 (NIC v65.74): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v8.30(8.33) (NIC v65.74): Full Scale 'Everest' *WHQL*
v9.92 (NIC v65.74): Full Scale 'Everest' *WHQL*
v9.95 (NIC v65.74): Full Scale 'Everest' *WHQL* Stability: HD Tach 3.0.4.0 (Done by v65.65 NIC)
v9.98 (NIC v65.74): Full Scale 'Everest' *WHQL* Stability: 'Everest' Buffered Reading (Done by v67.72 NIC)
v9.99.04-[Modded by Fernando] (NIC v65.74): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 (Done by v65.74 NIC)
v9.99.09 (NIC v65.74): Full Scale 'Everest' *WHQL* [Done at 790-SLI board]
v9.99.09-[Modded by Fernando] (NIC v65.74): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v9.99.09-[Modded by Fernando] (NIC v67.76.0-Vista version): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v10.1.0.14 (NIC v65.74): 'Everest' Linear - Buffered Stability: HD Tach v3.0.4.0 (Done by v65.65 NIC)
v10.1.0.20: No full scale available *WHQL* Stability: 'Everest' Buffered (Done by v65.65 NIC)
v10.3.0.16 (NIC v65.74): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0 (Done by v65.65 NIC)
v10.3.0.21 (NIC v67.72): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0
v10.3.0.21 (NIC v65.55): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0
v10.3.0.21-[Modded by Fernando] (NIC v67.76.0): Full Scale 'Everest' Stability: HD Tach v3.0.4.0 - HD Tune v3.00

[Extra] Windows XP Drivers at Windows Vista system:
v6.66-(NIC used was most stabile one v67.76.0): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v6.86-(NIC used was most stabile one v67.76.0): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0 - HD Tune v3.00
v6.99-(NIC used was most stabile one v67.76.0): Full Scale 'Everest' *WHQL* Stability: HD Tach v3.0.4.0 - HD Tune v3.00


General Notes:
Standard IDE controller [MS]: Finally managed to get this driver from hiding. You can get this, but you'll have to remove the driver store for MS v8.xx drivers & any other driver for it to stick in system also WUA(Windows update) will update it over first sign it sees it. However, amazing thing is this supports S.M.A.R.T. fully on Vista x86/x64 systems. Also seems to kick some speed in 208MB/s ain't bad for tops, but no good with out bad. These are incredible unstabile drivers I wouldn't be surprise, if something would literally shutdown because of spin downs on these. So for tweaking these defrag defrag defrag.. Doesn't feel open as in XP world and even in smallest software run these will have incredible slow downs.
(Read caching(Disabled by default) write/read caching can be found inside device properties, NCQ (No options - unknown), S.M.A.R.T. Supported & enabled, No External nor internal HDD corrected naming detection) (Uses driver files: 'atapi.sys', 'ataport.sys', 'pciide.sys', 'pciidex.sys') (Almost couldn't believe it myself that there is an driver which would suppport smart so here's prove of it: S.M.A.R.T. Supported)

v8.30(v8.22) -> Very unstabile driver from Microsoft bottom of ATA drivers. However this is still fast driver to consider being the end.
(Vista x64 only: If any versions of Nvidia Sata drivers has been loaded with the system. This driver still loads on background unless you overwrite 'nvstor.sys' at your 'windows\system32\drivers' folder and YES this does slow down new drivers)

v8.30(v8.33) -> Very unstabile driver from Microsoft bottom of drivers at Service pack 1. However this is still fast driver to consider beeing the new end.
(This was also secret update by Microsoft Update) (Vista x64 only: If any versions of Nvidia Sata drivers has been loaded with the system. This driver still loads on background unless you overwrite 'nvstor.sys' at your 'windows\system32\drivers' folder and YES this does slow down new drivers)

v9.92 -> Introduces NcQ to Vista to chipsets NF 4,4x0,5x0 & 6x0x series had v9.95 at this time having NcQ disabled through those. Very very stabile driver for system also supports more drives now detected correctly.
(The Advanced Caching is defaulted off. NcQ is defaulted ON. Vista x64 only: Leaves the 'nvstor.sys' as v8.30 you have to manually overwrite it)

v9.95 -> Again even very stabile driver out from Nvidia and supported by all chipsets 'except new 7x0 series ofcourse. This is again completly new driver. has no NcQ or S.M.A.R.T. Support. NcQ is default disabled & greyed out 'only manual registry editing can enable it.
(The Advanced Caching is defaulted off (If any earlier drivers like v9.92 had been installed it's defaulted as previous driver determed). NcQ is defaulted OFF & Greyed on NF4,4x0,5x0 boards.) (Vista x64 only: Leaves the 'nvstor.sys' as v8.30 you have to manually overwrite it)

v9.98 -> This driver game with a reason little later it's quite a bit faster and it's stabile as a rock. This was kinda compination driver of v9.95 with improved HardwareID support to support all chipsets NF4,4x0,5x0,6x0. This is something really nice to have on system. No S.M.A.R.T. or NcQ support on anything lower than NF 6x0 board.
(The Advanced Caching is defaulted off (This driver turns this feature off no matter what is the before driver). NcQ is defaulted OFF & Greyed on NF4,4x0,5x0 boards.) (Vista x64 only: Leaves the 'nvstor.sys' as v8.30 you have to manually overwrite it)

v9.99.04 -> Was now tested. As previously mentioned not tested. These are as fast but a stable model of v10.3.0.16 meaning there is no issues on Vista with these what so ever except the missing (v9.99.09 & v10.3.0.21) HotSwap support. These are not official except for RAID systems and probably will give you error 'not valid to this system' because of it you have to manually update the .sys files > reboot > then try install it again.
(These drivers will be at 'Storage Controllers' Section at your 'Device Manager' so don't be suprise it is still correct place since they are microsoft made, hehe.) (Vista x64 only: Leaves the 'nvstor.sys' as v8.30 you have to manually overwrite it)

v9.99.09 -> Newest on the series and follows the v9.99.04 now on bug fixed & has very very stabile interface & Fast. No S.M.A.R.T. or NcQ Support. Newly came Nforce 7x0 series is now supported on this driver and this driver also works as true Hotswap like v10.3.0.21 just alot faster. These also on some boards needs v9.99.04 style installation.
(The Advanced Caching is defaulted off (If any earlier drivers like had been installed it's defaulted as previous driver determed). NcQ is defaulted OFF & Grey'ed on all boards.) (Vista x64 only: Leaves the 'nvstor.sys' as v8.30 you have to manually overwrite it.)

v10.1.0.xx -> Has bad spin down problems. I don't really know what system these series were designed, but has incredible bad stability & bugs all the time from NF4-6. They are although quite fast considering v9.95 or older & quite usable when though other option was v9.92 at this time.
(The Advanced Caching is defaulted off (If any earlier drivers like had been installed it's defaulted as previous driver determed). Features Vista NCQ Support. No S.M.A.R.T.) (These are build to 'nvrd.sys' & 'nvrd64.sys' so think these were optimized for RAID design systems. However, there is existing WHQL 'nvstor.sys' and 'nvstor64.sys' also. Vista x64 only: Leaves 'nvraid.sys' to system to be manually overwritten & leaves 'nvstor.sys' to be manually overwritten.)
*Update* These drivers will be re viewed at some point. The versions 10.1.0.26 for vista is perfectly stabile including NcQ support this is like another side of v10.3.0.21 of AHCI which again includes Hot-Plug & new Acoustic (grey'ed), but not NcQ Supports.

v10.3.0.16 -> Detection failure after ~24hours of use ones. Reboots System cold with no bluescreens.
(The Advanced Caching is defaulted off (If any earlier drivers like had been installed it's defaulted as previous driver determed). No S.M.A.R.T. or NcQ Support. Made to NF410,430 & NF7x0 series however fully functional on all Chipsets. Includes completly new features greyed in anything less than Nforce 7x0 chipsets called 'command completion coaslescing', 'AAM' "0-255 values"(Automatic Acoustic Management) & 'Free fall control' (I do not know what these are don't ask).) (Functions with 'nvstor.sys' or 'nvstor64.sys'. Vista x64 only: Leaves behind v8.30 nvstor.sys at x64 systems Still loading both by default.)

v10.3.0.21 -> This is the Fixed version of v10.3.0.16/20 HotSwap support. This does work however is a lot slower.
(The Advanced Caching is defaulted off (If any earlier drivers like had been installed it's defaulted as previous driver determed). No S.M.A.R.T. or NcQ Support. The Faulty Reboot on v10.3.0.16 was also fixed these drivers are slow, but stabile in this case. Made to NF410,430 & NF7x0 series however fully functional on all Chipsets. Includes completly new features greyed in anything less than Nforce 7x0 chipsets called 'command completion coaslescing', 'AAM' "0-255 values"(Automatic Acoustic Management) & 'Free fall control' (I do not know what these are don't ask)) (Functions with 'nvstor.sys' or 'nvstor64.sys'. Vista x64 only: Leaves behind v8.xx nvstor.sys at x64 systems Still loading both by default.)

NOTE: Drivers uses 'nvrd32.sys' & 'nvrd64.sys' depending on system architecture for RAID including the above mentioned SataIDE sys file(s). (Means it still load windows default 'nvraid.sys' v8.xx(or updated from version) which should be removed to prevent false double loading and version mixups).


Extra Notes of Windows XP Drivers at Windows Vista System:

v6.66-[XP x64 Drivers under Vista x64]: no lag on boot up. These are quite stabile to consider that v6.99 are very very unstabile compared to these however, these are as slow or slower than v6.99. Doesn't seem better than the 'Standart IDE Controller', however seems to have huge load, speed issues without NcQ enabled. Has also CPU usage issues can't be stabile even idle. Check benchmarks notice the stability & HD Tune/tach burst speeds they still are near best 205MB/s drivers atm on vista with NcQ these comes very very close, but as it doesn't help any on load as in system load or database/games loading it kinda loses it's purpose. Drive Detection is missing as in XP the Harddrive name isn't printed full. All in all still seems very very well functioning drivers on Vista x86 and x64, if you don't mind speed considering having all features except HotPlug.
(NCQ is available & enabled, Cache Write/read Available (at controller) enabled by default, S.M.A.R.T. is functional) (Use Driver: 'nvata.sys' or 'nvata64.sys')

v6.86-[XP x64 Drivers under Vista x64]: No lags on boot up. These are vry stabile drivers comparing to either of v6.66 nor v6.99 and rest of the v6.xx series. driver is still as slow as the v6.66 or v6.99. Doesn't seem to be any better that 'Standard IDE Controller', has still huge load issues without NcQ enabled. These can be stabile doesn't seem to have exact same issue with CPU as v6.66 ot v6.99 had for somereason. Check more information on benchmarking. speeds are near same as v6.66 alot better disk read sounds/acoustic than v6.99. and again NcQ doesn't help any on load as in system load or database/games loading it kinda loses it's whole purpose. Drive Detection is still faulty (seems whole serie issue) does detect only paritial HDD naming and these as v6.66 seems very nicely functioning in general at Vista x86 and x64 systems still without hotplug.
(NCQ is available & enabled, Cache Write/read Available (at controller) enabled by default, S.M.A.R.T. is functional) (Use Driver: 'nvata.sys' or 'nvata64.sys')

v6.99-[XP x64 Drivers under Vista x64]: Near no lag on boot-up loading, Doesn't seem better than the 'Standart IDE Controller', however seems to have huge load, speed & stability issues without NcQ enabled. Has also CPU usage issues seems harder than hell to get these drivers stabile even on idle mode anyhow check benchmarks they are as usual done as in complete idle mode notice the HD Tune/tach burst speeds they still are near best drivers atm on vista with NcQ these comes very very close to be best, but as it doesn't help any on load as in system load or database/games loading it kinda loses it's meaning. Drive Detection is missing as in XP the Harddrive name isn't printed full. All in all still seems very very well functioning drivers on Vista x86 and x64, if you don't mind speed considering having all features except HotPlug, but these has to be compared to v10.1.0.xx driver series which is actual Vista simlar support drivers for speed.
(NCQ is available & enabled, Cache Write/read Available (at controller) enabled by default, S.M.A.R.T. is functional) (Use Driver: 'nvata.sys' or 'nvata64.sys')

XP drivers in Vista NOTE: Removing these drivers is not possible, if you remove the driver from your 'device manager' and delete driver files your Vista won't start anylonger because it leaves 'nvata' entry as loading which after that action doesn't have any signatures. There is only few solutions to this. Here's couple. copy WHQL sugned 'nvstor.sys' over 'system32\drivers\nvata.sys' or 'nvstor64.sys' to 'system32\drivers\nvata64.sys' will let you boot up and after that remove the nvataxx entries from driver load with tool like 'Autoruns' update the driver back to normal vista driver. Notice that restore points cannot be used nor 'last known good logins' it will still be there. Only other option for this is to update over, but won't suggest that unless using cleanup of somesort for nvataxx driver because it will generate the error then also. Also should be taken under consider that anything these drivers writes in Vista machines can be corruted when you update these to 'next' drivers you are going to use. The writing is incorrect & will generate/report error by update -> chkdsk. So be sure you accept this before even considering to install these. Also going to include screen shot here because when you update to next drivers (or try to remove) no matter what they are your system won't boot up because it leaves trace of v6.xx drivers to driver load which will not start because then it has no signs by microsoft. So you have to MANUALLY remove it after update before reboot to it to go smooth. Here is the Picture.
*Extra future Information* There is coming an 'optional.reg' file included on 'Fernando 1' Drievr packs in future which can manage the clean up restore the good none duplicated load state of Sata Performance drivers this reg file most be used AFTER the update back to new Vista Sata drivers.

RAID NOTE: v6.xx XP->Vista drivers uses the 'nvraid.sys' for x86 Vista & 'nvrdx64.sys' for x64 Vista when installed for raid systems. (On x64 side this is not default and will create a bit chaos. so on x64 remove the 'nvraid.sys' from loading if updating to these to get rid of duplicated loading & less mixup on loading).


Windows Vista Programs, caching & system set:
Every single benchmark and testing were done in same machine. All Vista basic software after installation were on. and no other extra software were installed on any state exception of 'Everest' for making Benchmarks. Rest of the drivers were updated so:
-Ethernet Controller: *Update* NIC used on Benchmark Notified with the benchmark.
-SMBus: v4.6.3
Other:
-No other devices or drivers were on system or by detected by Vista.
-No other software were run in the testing
-Windows Vista Caching aka. Superfecth were filled full before any Benchmarks were began to it to not interfere the result & for get the best usage result from system.
-Drivers were installed hardcore way. System was stripped down always to v8.30(v8.33) MS drivers after each benchmark & Driver files were deleted before new drivers were installed to make sure there is no left overs at system.

[Fernando 1]

Hello TwL!

On behalf of all nForcersHQ forum members I want to thank you for the competent and very informative nForce S-ATA driver review and the detailed comparison of the different versions.
Your benchmark results and your specific recommendations are a great help for all nForce chipset mainboard owners, who are searching for the best possible nForce IDE drivers depending on their specific requirements.

Without your assistance giving me frequently your test results and recommendations I would not have been able to present and optimize my special NF4-7 chipset driver packages for Vista and Windows XP (32/64bit). So I think, that our cooperation is very useful for all nForce chipset users.

As a consequence of your test results I have already modified the INF files of the brandnew Vista and XP 9.99.0.9 IDE drivers to customize them for all nForce chipsets from NF4 up.
Within the next hours I will mix and upload new NF4-7 chipset driver packages for Vista and Windows XP with the following components:

[TwL]

@Fernando 1

Looks fine to me. Although, I am atm testing the new '67.xx' series network drivers possible impact/affect to Hard drive working on Vista. At XP that 67.72 driver makes huge different bad ways how the hard drives works as you can see those benchmarks couldn't be made even with it it had like 10-30MB/s Buffered reading difference depending specially what network controller driver you use. So I assume this same matter also is in Vista.

I think the new network drivers has something to do with transformin averate/normal hard drives to be closer as network hard drives to make some sort of new style of identifying hard drives to systems like active directory usage or perhaps speed up "common gateway"-folders with multible PCs and that way has kinda slow down since it checks is something there or not. On XP technology this could very well be the reason why there is lag on new drivers because it would have to be loaded/checked on start-up.

[Fernando 1]

[TwL]

@fernando 1

Not really sorry mate..

Since on XP world it's like lower you go better it runs(Less it stress boot-up & usage), hehe. NF5-6 boards Networking controller is as low as v60.xx to be working good while those has a bit lag on v65.xx, but even that is better option than the new ones and that v65.xx is again official NF4 driver. and none of these are compatible with new chipse except the v67.xx series.

So can't much recommend any single network driver for XP unless I say NF4-6 v65.xx and NF7 v67.xx.

[Fernando 1]

[TwL]

[Fernando 1]