kondilac

Actually he stated in his first post, and also visible on the 3dm results that he has the iGPU enabled, but also theNV5200, and i think the presence of the dedicated GPU could be the issue why the optimus compression is not kicking in, but i'm really just tapping in the darkness bc i never tried a laptop with dgpu + egpu.

You should install the Dell Feature Enhancement Pack, especially the component responsible for power profiles. But as far i'm concerned, the only benefit of this is that you can set the fan to max levels, which is +500 RPM compared to the available without it. The power profiles provided by this pack also control the CPU throttling mechanism of the embedded controller, but i don't really think that the problem is caused by the embedded controller, nor the power profiles. If you are getting lower GPU score with the same OS, same connection, same drivers, the only thing comes to my mind is that the pcie compression provided by nVidia Optimus is not enabled for some reason, and i guess the dGPU is responsible for that, but since i own the iGPU only model, i can not help you from this point.

The effect of higher physics score on 3dm points is very small, upgrading from a i5-3230M to i7-3840qm yielded ~300p better scores, but the physics score increase was more than 120%. If you still have access to the x230, maybe you should check the graphics settings in the nVidia control panel, and make sure that the settings are the same on the e6430 setup.

Not so long ago, i have kind of finished upgrading the hardware in my 6430. I can only tell one thing, this machine is a beast! 3dM11 results With a slightly overclocked GTX560Ti @1.2Opt The memory controller of this i7-3840QM can run my RAM sticks at factory timings @2133 with no problem, althrough only the slightest BCLK raise freezes the machine. No problem, i am fine @4Ghz 4C/8T The performance is on par with the desktop Ivy i7-s! I do not use C3 &C6, but turn on C1E to reduce temps & power consumption occasionally. I made aproppriate Throttlestop profiles for my usage scenarios. Idling power consumption is around 6W, but @full load the CPU is over 55W. Since i have the iGPU only model, the hetsink is inadequate, so currently i am looking for a dGPU heatsink to mod to fit. I use a Dell PR03X dock (with usb3) and i created a neat little script that automatically changes TS profiles when docked / undocked, and... As i found out, my current eGPU setup(W8.1 x64 UEFI, x1.2Opt @EC) can be hot-plugged on the go. No need to sleep / wake, i can just plug it in, and it instantly activates, aswell my 2 displays connected to the GPU. The other thing that i noticed, that i can "Safely Eject" it from the taskbar menu (disabling from device manager also works, but has to be enabled manually to regain true hot-plug) , and then remove the adapter without any system crash happening. So i came up with a very convienent solution wich fits my workflow when docking the laptop: I've made a script, which runs in the background, and detects if the machine is undocked, and automatically ejects the GPU. So basically, i can dock the laptop, plug in the EC, and everything works flawless, then without any interaction in the OS. When i take the laptop i only eject the machine from the dock, and can unplug the EC without any problems, without the need to restart/ sleep/wake the machine. The only interaction still required is to manually switch on and off the ATX PSU which feeds the GPU with juice, but this could be easily automated using the "dell monitor stand connector" which indicates when the laptop is connected. Work in progress... The script is kind of in beta state right now, but anybody who is interested hit me up with a PM and i send it, with the minimal instructions currently needed to configure it. Oh yeah and it does not need a docking station, theoretically i can bind the GPU eject to the removal of any hardware, (USB hub,mice,thumbdrive). Of course it cant be connected through the EC-s USB port because the removal of the EC must be after the removal of the helper device.

I really like that carbon-effect foil on the e6430. DIY?

OK, so still no success overclocking BCLK. I have experimented a lot, bricked the machine a dozen times. I will share my experiences, but basically it will be a wall of text regarding how buggy the dell BIOS update mechanism is, there is something fishy with the contents of my flash, and that i tried every possible combination of bios version, modifications, ME FW, but the best i could get is a 0.5% slider, which has no practical effect (and it is the same as i got the machine). Before everything, could You please upload me a complete dump of Your flash contents with FPTW? My last hope is to check how does that work out if i use a full copy of the contents of your flash.

I checked the one provided by Khenglish, and my dump side by side, but have not identified any differences. maybe i was too tired. I have also sucessfuly flashed the one provided by Khenglish, but then there isn't even a slider Of course i followed the instructuons so cold started the machine, reinstalled (the newest) XTU but no slider. The dump, wich has slider but bad range Interestingly, i have "updated" from A16 to A16 just bc curiousity, and compared the vanilla bios region with the dump from my flash, and they are not identical.

So it turns out my ME firmware was already modded, by the previous owner i guess, with the exact same settings as the one provided by Khenglish. Too bad, that the slider has a range of 99.5-100, and does nothing in practice, i guess because my CPU is only an i5. Interesting that the decsriptor was locked back. RAM is low priority, because i got it stable at 2133@cl11 CR2 (still no reset tho but i haven't actually tried anything since), but i tinkered with the BIOS a bit, and there is no documented way to unlock the hidden menus. Its an AMI Aptio, customized by Dell. At least i can be sure that advanced settings are only hidden from the user, not that DELL does not even implement them in the setup module. I will try my best but i have not much time on weekdays. In the worst case i could get a tree-like structure with all the options, and info, wich is more clean then the raw extracted var list currently provided by timohour, but essentially contains the same info already available.. What i have discovered in the last days is that the flash descriptor doesn't get locked back by a bios update, and the BIOS region write protection can be sucessfully disabled by setting EFI var 0x40 to 0x0. I read somewhere in this thread that there were some problems regarding this. Also my HDD caddy arrived (3.5$ from eBay), and i'm very happy that the latch/ejector thingy can be placed on it from the DVD, along with the faceplate. So basically the cheapest available 9.5mm caddy from eBay is as good as the more expensive ones. One little difficulty is that the 2 fixing screws are too short (At least fot this caddy) and has to be replaced. I had compatible screws, but anyone who doesn't can buy them from an optician shop for example.

I'm having trouble running my Kingston HyperX 16Gb 2133Mhz/CL11 kit with my Ivy Bridge Mobile CPU.(i5-3230M). The laptop is a Dell e6430. Using auto-memory configuration settings results in an unusable system; cant even manage to get to POST. Manually setting the freq. to 1866 works like a charm, with the factory timings. Manually setting the freq to 2133, and the command rate to 2T boots up, and works flawless... ...until i try to restart the system: I only get black screen after a restart, and cant "cold boot" again even after removing both modules, AC, and the battery for a short time. After removing everything for at least half an hour, the system boots up again, and works without issues(@2133 CL11 CR2, factory timings). I can shutdown the system, and it is able to start again without problems. Only restarting introduces the problems. The available settings in BIOS: <code id="yui_3_10_0_1_1447716287936_1846" class="php" data-component="highlight">DIMM profile Variable: 0x1EE Options: Default DIMM profile Value: 0x0 Custom profile Value: 0x1 XMP profile 1 Value: 0x2 XMP profile 2 Value: 0x3 DDR (voltage) Selection Variable: 0x1E9 Options: DDR3 Value: 0x0 DDR3L Value: 0x1 Auto Value: 0x2 DDR Reset Wait Time Variable: 0x1EA 32 Bit Numeric Value Memory Frequency (DIMM1) Variable: 0x1E6 Options: Auto Value: 0x0 1067Mhz Value: 0x1 1333Mhz Value: 0x2 1600Mhz Value: 0x3 1867Mhz Value: 0x4 2133Mhz Value: 0x5 2400Mhz Value: 0x6 2667Mhz Value: 0x7 Memory Frequency (DIMM2) Variable: 0x1E7 Options: 1067Mhz Value: 0x42B 1333Mhz Value: 0x535 1600Mhz Value: 0x640 1867Mhz Value: 0x74B 2133Mhz Value: 0x855 2400Mhz Value: 0x960 2667Mhz Value: 0xA75 ECC Support Variable: 0x1DD Options: Disabled Value: 0x0 Enabled Value: 0x1 Max TOLUD Info: Sets the maximum value of the Top Of Low Usable DRAM (TOLUD) setting, determining how much space is allocated for PCI devices below 4 GB. Dynamic assignment adjusts TOLUD automatically based on the largest MMIO length of the installed graphics controller. Variable: 0x1E Options: Dynamic Value: 0x0 1 GB Value: 0x3 1.25 GB Value: 0x4 1.5 GB Value: 0x5 1.75 GB Value: 0x6 2 GB Value: 0x7 2.25 GB Value: 0x8 2.5 GB Value: 0x9 2.75 GB Value: 0xA 3 GB Value: 0xB 3.25 GB Value: 0xC NMode Support Info: Command rate. The number of rising edges that must be present between the assertion of the Chip Select commend and the first Row Active command. In a 1T configuration the RA command can occur on the cycle immediately following CS. In a 2T configuration, a no-op must be inserted for one cycle. Variable: 0x1EF Options: Auto Value: 0x1 1N Mode Value: 0x0 2N Mode Value: 0x2 Memory Scrambler Info: Variable: 0x1F0 Default: Enabled Options: Enabled Value: 0x1 Disabled Value: 0x0 RMT Crosser Support Variable: 0x1F1 Default: Disabled Options: Enabled Value: 0x1 Disabled Value: 0x0 MRC Fast Boot Info: Enable Memory Fast Boot to skip DRAM memory training for faster booting. Variable: 0x1DE Default: Enabled Options: Enabled Value: 0x1 Disabled Value: 0x0 Force Cold Reset Info: Force cold reset or choose MRC cold reset mode, when cold boot is required during MRC execution. Note: If ME 5.0MB is present, Force cold reset is required! Variable: 0x1DF Default: Enabled Options: Enabled Value: 0x1 Disabled Value: 0x0 Scrambler Seed Generation Off Info: Variable: 0x1E0 Default: Disabled Options: Enabled Value: 0x1 Disabled Value: 0x0 Memory Remap Info: Enables or disables Memory Remap above 4 GB, allowing more space for 32-bit PCI MMIO Variable: 0x1E1 Default: Enabled Options: Enabled Value: 0x1 Disabled Value: 0x0 Memory Alias Check Info: Variable: 0x1E2 Default: Disabled Options: Enabled Value: 0x1 Disabled Value: 0x0 DIMM Exit Mode Info: Sets the DIMM Exit Mode. In Slow Exit mode, the memory clock DLL will be powered down in S3 suspend mode. Variable: 0x1F2 Default: Fast Exit Options: Auto Value: 0xFF Slow Exit Value: 0x0 Fast Exit Value: 0x1 Power Down Mode Sub-settings Info: Sets the power down mode. When DIMM Exit Mode is configured to Slow Exit mode Variable: 0x1F3 Options: DLL-off Value: 0x6 APD-DLLoff Value: 0x7 When DIMM Exit Mode is configured to Fast Exit mode Variable: 0x1F4 Default: PPD Options: No Power Down Value: 0x0 APD Value: 0x1 PPD Value: 0x2 APD-PPD Value: 0x3 Memory Thermal Configuration Memory Thermal Management Variable: 0x1F9 Options: Disabled Value: 0x0 Enabled Value: 0x1 PECI Injected Temperature Variable: 0x1FA Options: Disabled Value: 0x0 Enabled Value: 0x1 EXTTS# via TS-on-Board Variable: 0x1FB Options: Disabled Value: 0x0 Enabled Value: 0x1 EXTTS# via TS-on-DIMM Variable: 0x1FC Options: Disabled Value: 0x0 Enabled Value: 0x1 Virtual Temperature Sensor (VTS) Variable: 0x1FD Options: Disabled Value: 0x0 Enabled Value: 0x1 Memory Timings tCL Variable: 0x1FE 8 Bit Numeric Value tRCD Variable: 0x1FF 8 Bit Numeric Value tRP Variable: 0x200 8 Bit Numeric Value tRAS Variable: 0x201 16 Bit Numeric Value tWR Variable: 0x203 8 Bit Numeric Value tRFC Variable: 0x204 16 Bit Numeric Value tRRD Variable: 0x206 8 Bit Numeric Value tWTR Variable: 0x207 8 Bit Numeric Value tRTP Variable: 0x208 8 Bit Numeric Value tRC Variable: 0x20 16 Bit Numeric Value tFAW Variable: 0x20B 16 Bit Numeric Value</code> Can anybody enlighten me, how fix this reset issue? My first guess is extending the "DDR Reset Wait Time" but before experimenting on my own, i tought i ask advice from experts on the topic. Other users with the same machine(with better i7 quadcores), and same, or previous generation HyperX kit who are running@ 2133 (and with BCLK OC even @2340) are reporting that they had to loosen tCL to 12, but CL11 is still performing better with 2T command rate, than CL12 1T , and obviously i'm after the best performance possible. Thanks in advance for the help.

Of course, I intended that post to be embedded in Your first post Thanks for the advices. I have a hard time getting accustomed to this board's formatting options, that's why i embedded it as "code". I wrote that post in text editor, and indented it to be easy on the eye but when copy/pasting it into the text editor of course the indention got messed up. I will continue to update that post in the next few days with additional info, as i'm experimenting with memory settings.

Finally, i reached "adult" status on this board, and now i can post by myself I successfully upgraded the SB i3, to an IB i5-3230M for 15$. Since my HyperX 2133Mhz kit is already migrated to the Dell, the first boot revealed that the memory controller of the i5 cant cope with the PnP (JEDEC) 2133Mhz @CL11. I locked the memory speed down to 1600Mhz CL11 , replacing one of the kingstons with a samsung module, just to be able to check the new CPU is working properly. Yes it does, and a little bit of TSBench revealed that it does outperform my Haswell-ULV on the same clock not only with lower power consumption (15.1W thanks to the Haswells IVR and integrated PCH VS 10.9W) but a 32M TSBench completes almost a second faster. The Haswell system has a very lean w8.1 x64 install, and the Ivy has a bloated w7 x64. (no throttling occurs, both cpus are running @2.4Ghz during the whole bench)The other thing which makes me very happy that when C-states are disabled in the bios, i can still toggle C1E using Throttlestop for the Battery profile(I was able to do the same on 1st gen. Core i anno, but unable to do it with 4th.gen.), and although it does not enable C3, C6 and C7, but still has a substantial impact in idle power consumption (3.1W vs 7.1W). And the advantages of disabled C states in my usage scenario is huge, so the ~+1W power consumption when idling is a fair trade-off for me.(This i5 idles around 2.5W with all C states enabled, although i have not tinkered with the settings so far to see if this can be reduced)The next thing on my list is finding the limit of the memory controller, but i'll have to dig into the topic deeper... My kit is a Kingston HX321LS11IB2K2/16, if anybody with the same kit would share their settings that would be great.UPDATE: Using EFI variables (Thanks timohour for the extract of A16 BIOS), i was able to lock the memory freq. to 1866, so my kit is running in dual channel 1866Mhz 10-11-12-30 CR1. This could be useful for some other folks too, when the automatic freq. configuration makes the system unusable, but one can get into the one-time boot menu (F12 during Dell logo). Unfortunately my machine wont even get to POST when the auto memory frequency configures the sticks to run @ 2133. Also, changing boot mode from Legacy to UEFI when one wants to mangle with EFI vars is unnecessary, because even in Legacy mode there is option to boot into the EFI shell from the USB drive. I guess SecureBoot still has to be disabled, but i will check. Setting: Memory Frequency, Variable: 0x1E6 {05 A6 12 04 13 04 D1 01 02 00 E6 01 10 10 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00}0x60BA3 Option: Auto, Value: 0x0 {09 0E BE 02 30 00 00 00 00 00 00 00 00 00}0x60BB1 Option: 1067, Value: 0x1 {09 0E 14 04 00 00 01 00 00 00 00 00 00 00}0x60BBF Option: 1333, Value: 0x2 {09 0E 15 04 00 00 02 00 00 00 00 00 00 00}0x60BCD Option: 1600, Value: 0x3 {09 0E 16 04 00 00 03 00 00 00 00 00 00 00}0x60BDB Option: 1867, Value: 0x4 {09 0E 17 04 00 00 04 00 00 00 00 00 00 00}0x60BE9 Option: 2133, Value: 0x5 {09 0E 18 04 00 00 05 00 00 00 00 00 00 00}0x60BF7 Option: 2400, Value: 0x6 {09 0E 19 04 00 00 06 00 00 00 00 00 00 00}0x60C05 Option: 2667, Value: 0x7 {09 0E 1A 04 00 00 07 00 00 00 00 00 00 00} There is another EFI var that for DIMM2 (the top slot) but it does not need to be modified if running dual chanel, because the freq, and timings will be determined by the module with the worst capabilities. I've also noticed that there is something wrong with the values, because this should be only 8 bit, instead of 12 (?!)Setting: Memory Frequency, Variable: 0x1E7 {05 A6 12 04 13 04 D2 01 02 00 E7 01 10 11 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00}0x60C3B Option: 1067, Value: 0x42B {09 0E 14 04 00 01 2B 04 00 00 00 00 00 00}0x60C49 Option: 1333, Value: 0x535 {09 0E 15 04 30 01 35 05 00 00 00 00 00 00}0x60C57 Option: 1600, Value: 0x640 {09 0E 16 04 00 01 40 06 00 00 00 00 00 00}0x60C65 Option: 1867, Value: 0x74B {09 0E 17 04 00 01 4B 07 00 00 00 00 00 00}0x60C73 Option: 2133, Value: 0x855 {09 0E 18 04 00 01 55 08 00 00 00 00 00 00}0x60C81 Option: 2400, Value: 0x960 {09 0E 19 04 00 01 60 09 00 00 00 00 00 00}0x60C8F Option: 2667, Value: 0xA75 {09 0E 1A 04 00 01 75 0A 00 00 00 00 00 00} UPDATE2: I was also able to boot into windows @2133 CL11-12-13(factory) , when manually setting EFI var 0x1EF (Nmode) to 2N instead of auto. Using 2N mode improves stability, but 1N offers the maximum performance. Memory bandwith R/W/C was around 28k/30k/28k Mb/s. Before running a GFX bench to check for iGPU instability, i've restarted the system. Tried to restart, to be exact, because this functionality is lost, but there are some other settings available in the bios at the memory section, so maybe this could work out better than loosening the timings. I will continue to experiment tomorrow, until then i've reverted back to 1866, which is stable with the factory (tight) timings.

Single and dual channel benches @ 2133 CL11 CR2 Also it turns out that after a shutdown, there is no problem booting up, but restarting causes black screen, and problems even when trying to coldstart. iGPU has no artifacts, and everything is fine once i can get it to boot. AFAIK none of the i5 bear with this capability, setting the 1C and 2C max multis to 33 (0x21) had no effect. (as visible on the benches above) Yes, iGPU is overclockable in this i5. Although enabling GT OC Support does not enable manual tuning the clock and voltage in XTU (like with Haswell) but raising the freq. in the BIOS works. Interestingly modifying these settings resulted in that previously available turbo power settings are now greyed out in XTU, will investigate this issue further. Also i will try to raise the BCLK (by default, i can set it between 99.5 and 100, but it looks like it doesn't affect BCLK at all, as visible in TS), but i have to retrieve my SPI programmer, because i really don't want to mangle with the pinmodding method. Results will come next week. UPDATE: The following post by me is my fist (but not last) contribution to the info-base in the first post of this thread. It contains all the memory configuration settings available in the BIOS, with EFI vars, and occasionally some extra info about the setting, and default values. Fiddling with EFI vars is very inconvienent, especially with the lack of scripting capabilities (efi/boot/startup.nsh does not work), so it would be a real ease to somehow unlock all the hidden menus in the Aptio bios. I will check the possibilities of doing that, but even if it is possible, a hardware SPI programmer is necessarry AFAIK.

Finally, i reached "adult" status on this board, and now i can post by myself I successfully upgraded the SB i3, to an IB i5-3230M for 15$. Since my HyperX 2133Mhz kit is already migrated to the Dell, the first boot revealed that the memory controller of the i5 cant cope with the PnP (JEDEC) 2133Mhz @CL11. I locked the memory speed down to 1600Mhz CL11 , replacing one of the kingstons with a samsung module, just to be able to check the new CPU is working properly. Yes it does, and a little bit of TSBench revealed that it does outperform my Haswell-ULV on the same clock not only with lower power consumption (15.1W thanks to the Haswells IVR and integrated PCH VS 10.9W) but a 32M TSBench completes almost a second faster. The Haswell system has a very lean w8.1 x64 install, and the Ivy has a bloated w7 x64. (no throttling occurs, both cpus are running @2.4Ghz during the whole bench) The other thing which makes me very happy that when C-states are disabled in the bios, i can still toggle C1E using Throttlestop for the Battery profile(I was able to do the same on 1st gen. Core i anno, but unable to do it with 4th.gen.), and although it does not enable C3, C6 and C7, but still has a substantial impact in idle power consumption (3.1W vs 7.1W). And the advantages of disabled C states in my usage scenario is huge, so the ~+1W power consumption when idling is a fair trade-off for me.(This i5 idles around 2.5W with all C states enabled, although i have not tinkered with the settings so far to see if this can be reduced) The next thing on my list is finding the limit of the memory controller, but i'll have to dig into the topic deeper... My kit is a Kingston HX321LS11IB2K2/16, if anybody with the same kit would share their settings that would be great. UPDATE: Using EFI variables (Thanks timohour for the extract of A16 BIOS), i was able to lock the memory freq. to 1866, so my kit is running in dual channel 1866Mhz 10-11-12-30 CR1. This could be useful for some other folks too, when the automatic freq. configuration makes the system unusable, but one can get into the one-time boot menu (F12 during Dell logo). Unfortunately my machine wont even get to POST when the auto memory frequency configures the sticks to run @ 2133. Also, changing boot mode from Legacy to UEFI when one wants to mangle with EFI vars is unnecessary, because even in Legacy mode there is option to boot into the EFI shell from the USB drive. I guess SecureBoot still has to be disabled, but i will check. Setting: Memory Frequency, Variable: 0x1E6 {05 A6 12 04 13 04 D1 01 02 00 E6 01 10 10 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00} 0x60BA3 Option: Auto, Value: 0x0 {09 0E BE 02 30 00 00 00 00 00 00 00 00 00} 0x60BB1 Option: 1067, Value: 0x1 {09 0E 14 04 00 00 01 00 00 00 00 00 00 00} 0x60BBF Option: 1333, Value: 0x2 {09 0E 15 04 00 00 02 00 00 00 00 00 00 00} 0x60BCD Option: 1600, Value: 0x3 {09 0E 16 04 00 00 03 00 00 00 00 00 00 00} 0x60BDB Option: 1867, Value: 0x4 {09 0E 17 04 00 00 04 00 00 00 00 00 00 00} 0x60BE9 Option: 2133, Value: 0x5 {09 0E 18 04 00 00 05 00 00 00 00 00 00 00} 0x60BF7 Option: 2400, Value: 0x6 {09 0E 19 04 00 00 06 00 00 00 00 00 00 00} 0x60C05 Option: 2667, Value: 0x7 {09 0E 1A 04 00 00 07 00 00 00 00 00 00 00} There is another EFI var that for DIMM2 (the top slot) but it does not need to be modified if running dual chanel, because the freq, and timings will be determined by the module with the worst capabilities. I've also noticed that there is something wrong with the values, because this should be only 8 bit, instead of 12 (?!) Setting: Memory Frequency, Variable: 0x1E7 {05 A6 12 04 13 04 D2 01 02 00 E7 01 10 11 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00} 0x60C3B Option: 1067, Value: 0x42B {09 0E 14 04 00 01 2B 04 00 00 00 00 00 00} 0x60C49 Option: 1333, Value: 0x535 {09 0E 15 04 30 01 35 05 00 00 00 00 00 00} 0x60C57 Option: 1600, Value: 0x640 {09 0E 16 04 00 01 40 06 00 00 00 00 00 00} 0x60C65 Option: 1867, Value: 0x74B {09 0E 17 04 00 01 4B 07 00 00 00 00 00 00} 0x60C73 Option: 2133, Value: 0x855 {09 0E 18 04 00 01 55 08 00 00 00 00 00 00} 0x60C81 Option: 2400, Value: 0x960 {09 0E 19 04 00 01 60 09 00 00 00 00 00 00} 0x60C8F Option: 2667, Value: 0xA75 {09 0E 1A 04 00 01 75 0A 00 00 00 00 00 00} UPDATE2: I was also able to boot into windows @2133 CL11-12-13(factory) , when manually setting EFI var 0x1EF (Nmode) to 2N instead of auto. Using 2N mode improves stability, but 1N offers the maximum performance. Memory bandwith R/W/C was around 28k/30k/28k Mb/s. Before running a GFX bench to check for iGPU instability, i've restarted the system. Tried to restart, to be exact, because this functionality is lost, but there are some other settings available in the bios at the memory section, so maybe this could work out better than loosening the timings. I will continue to experiment tomorrow, until then i've reverted back to 1866, which is stable with the factory (tight) timings.

More cache is always better, because it will reduce cache misses. L3 cache is a magnitude faster than the RAM, both latency, and bandwidth wise. Although it has no big impact on average workloads (also regular 3D gaming), L3 cache can have a substantial impact on workstation-like loads, with many simultaneous processes. Switching a process to another core is a very expensive operation, and the L3 cache can make this faster. I.e virtualization/simulation (especially if you are running more virtual machines than the # of cores) or rendering. Of course everything depends on the actual software used. The question is whether this worth the extra $. I was not really aware of the huge price difference between a 37x0 and 38x0 , but i do a lot of virtualization (networking simulation), also multitrack audio editing/rendering. So the conclusion is that the benefit from larger L3 depends on the workload. This thread has at least one user with a 38x0qm, maybe he can help you if you are curious if the larger L3 does affect the performance of a specific task. EDIT: Ok i have to learn to use the reply buton... EDIT: Dusty english grammar

For some reason, i was in the belief that unlike SB, in the IB generation only the XM models have 8M (and believe me, +33% cache counts), but You're right, and the 38X0QM chips also have the increased amount. I have no intentions in buying an XM now. Yes i've already read through this whole thread before even buying the machine, and know that the heat will be the limiting factor not the chip's maximum freq. By the way let me thank again for all the information You and others shared on this forum because without it, i don't think i would have ever think about buying this Dell machine, and so far I really like it. @R.O.G : the SATA nerfing is also removed starting from BIOS version F.29 ( i.e SATAIII works flawless)

OK, looks like my reply to R.O.G will never get approved by the mods. Yes, i totally forgot that it could be the cause that it is an i3, even worse SB. @R.O.G : Check the latest 2560p BIOS (F.60) , since F.28 the ram freq. nerf is removed. I had no problems running @1600 with an i5 in my 2560p. @timohour: Thanks for the advices, everything is fine with the unit. And yes, I definitely want to upgrade to IB, I'm aiming for an XM chip because the extra cache. But i have to find a (relatively) cheap one, until then i will get an IB i5, just to be able to migrate on the machine from my haswell POC. Skylake is already here in hand with Christmas, so i hope to find a used XM chip under 300$ soon (I got a 400$ quote on a new 3940XM . I have the iGPU only model, but my eGPU parts are already on the way.

Hi! I just joined the club! (although maybe not for so long) My e6430 arrived yesterday. i3-2350, 4Gb, 320Gb, 2x60Wh T54FJ, very good condition, warranty. I live in Europe, and it had cost me my 2560p + 70$. The BIOS version is A16, and the memory clock sped is locked down to 1333Mhz, although the module is 1600Mhz. Do previous BIOS versions have the same restraint? I'm planning to upgrade (I already have a 16Gb 2133 HyperX kit in my other machine (and know about the problems too), and a Samsung 850 Pro 256Gb SSD, so basically a better CPU is missing only) but fist i would like to be sure that everything is OK with the machine, because it still has 1 month warranty. Could any long time user give advices, what to scrutinise?

Hello! I recently bought a 2560p, and I would like to revert back to bios version F01, but the file from here: http://forum.notebookreview.com/thre...20#post8574086 is set to private in mediafire. If anybody have a copy, could you e-mail it to me please? Thanks in advance!