juandante

It is more like 1.7-1.8 seconds of difference because my "best score" is below. This is still very high for a 4% increase in frequency, nevertheless, without all those maths, the benchmarks seems in agreement, so both our systems should be running OK. Yours is faster, mine a little bit less faster, this was not the case for XTU benchmark. The only real issue I see is my temps because I easily hit 104 degres mark. I will repaste when I have time. I wonder for the 3940xm. I agree that it is unfortunate that we can't overclock... On laptops, it is the maximum multiplier that the CPU will have access to, but the CPU frequency will not be locked at this value. OEMs for laptops force CPUs to dynamically reduce the frequency even if setting the maximum multiplier because of heat. If you want best benchmark scores, crank up the maximum multiplier, for best battery, crank it down. You must set it to the max before starting the benchmark. I think if you remove the setting, the default maximum multiplier will be those that Intel statted, but I always set it at the max to be sure. On a desktop computer, the CPU will not dynamically reduce its frequency, but the temps will rise as hell, I am at 30 degs, and when I lock the multiplier I go way up to 60 degs on my desktop PC. Side note : Throttlestop will not apply any settings if you don't turn it on with the "Turn On" button, the benchmark is independent from Throttlestop and will not turn anything on, it will just run the benchmark.

Your results are, somewhat, in agreement with my results (3820qm). You max out at 3.5 Ghz on 4 cores and I max out at 3.4 Ghz on 4 cores (normally 3.5 for me and 3.6 for you but Windows 10...), hence, you gain 2 seconds on the 8 threads benchmark. But, I still find 2 seconds a lot. I suppose that : 1/ You are running in dual channel, this is not my case, which should give you 1 second, even if this seems a lot simply for running in dual 2/ Obviously the 3840 is more powerfull than the 3820, giving you again 1 second, but this again seems a lot because the ratio of frequencies 3.5/3.4 this should give you only 0.5 seconds maximum of performance advantage (ratio of frequencies can be done because exactly same architechture and class of CPUs), and also, this is confirmed if I review benchmarks for example of the famous Passmark, the 3840 is only 4% faster than 3820. Anyway, we are in agreement for the Throttlestop, not as with the XTU benchmark Also, I see that your tempareture is maxed out at about 85°. Me I max out at 104°. How is this possible ? Did you use your laptop with case open ? I am confused. Your temperature at CPU rest is the same as mine, about 50°C, and 3.5 to 5W. Someone did a very good observation about the Spectre pacth. I am running with the latest BIOS/patched OS, maybe this explains a lot

What is your score with ThrottleStop ? I have 14:3 seconds best score with a 3820qm and there is almost NO throttling, it runs during the benchmark at full 3.4 Ghz (3.5 Ghz is the maximum turbo but it seems that CPUs will almost never reach this maximum value on a standard non-stripped Windows 10) PS: you don't need to remove XTU or other utilities to run the benchmark of Throttlestop, just be aware that you must use the "turn off" bottom middle button to turn off Throttlestop

It seems that your 3940xm is 50% more powerful than my 3820qm which is simply impossible. You have 599 points and I have 410. First, due to the known maximum power output of 40w of the 2570p it is impossible that you run with such results. Even in considering that I overclock my 3820 qm at the frequencies of the 3940xm (which is impossible because oc is impossible on the 2570p), I will still not get 50% more points and jump to 599. I seriously doubt you get such results.

You seem to have exceptionnal results for a XM cpu. I have a 3820 and cant have as good results as you, are you sure the CPU doesnt throttle as often ? First it will throttle quickly for the simple fact that the mobo of the 2570p doesnt output 55w but 40w so from a theoritical analisys the CPU should run at 80percent of its full potential Can you download ThrottleStop and post the benchmark results ? (or another simple program as Cinebench, or CPUz, or HWmonitor, etc)

Just googled the pictures of your thing, I would not be at all surprise that there exist the same adapter with an external power source, do your homework

I am using 16gb on my 2570p it is not an absolute laptop problem it is probably due to your specific configuration. You are searching for help but update your bios first of all. Except if you have a good reason not to do so. Second, I had quickbooting problem when running dual channel with the exact same symptoms but fixing it was as simple as cleaning the RAM slot. Last samsung is not a reference in RAM modules so try the sticks on another laptop to see if it is the sticks, but in my experience this is the very last thing that can be the cause.

I was planning to get a new laptop (an ultrabook) with an i5 8x50u (something 8th gen i5 like this, running at 4.5 Ghz on 1 core). I did new benchmarks Throttlestop with the 3820qm. Actually I was shocked, the 3820qm : 9seconds for 8 threads. the 8th gen CPU : 9 seconds for 8 threads. (thus, the Windows 10 and ultrabook was in a shopping mall so very bloated, so maybe one could hope to gain 1 second by removing all the bloatware) Note : the ultrabook was throttling but the 3820qm on my 2570 is running at constant 3.4 Ghz with absoluetly 0 throttling (maybe 50 mhz max) at about 85°C. So if you want performance and portability for cheap, the 2570p is still OK. Unfortunatly (or fortunatly, for their prices at least), the ultrabooks have other advantages like 4K screens or battery autonomy.

Yes I have the 3820qm and I can only imagine the 3840qm obviously more powerful. The 3820qm for my part is indead a real pleasure to use. I think the CPU is powerful but on Windows 10 for everyday use the GPU is also very good and gives a snappy feeling. The GPU is extremely quick (for what it is supposed to be), matter I have a Dell with a NVIDIA quadro 2000m running at 200fps in one game and the 3820qm gpu is running at 140fps. Really this impresses me because before an Intel integrated GPU would be crawling and lagging like hell. Having a good GPU and a powerful CPU makes those quad core very good. And also, they run almost at max turbo on the 2570p. I tested an external Wifi module (TV decoder) on the 2570p wifi slot and it was working. On the wwan slot I think 1/ there is black list and 2/ anyway the wwan module connections are odd. I really hate HP for blacklisting their slots, this should be forbidden and is not a good act. Especially when considering the initial price of those laptops. Dell and Acer are the few that don't blacklist their slots. Yes I read also the Ivy bridge (any one) are limited to 32gb for quadcore and dualcore. For the Sandy bridge I CONFIRM that 32gb max is only for quad core and 16gb max for dual core at least with 8gb modules. I think it is a good try. If it doesn't work, just return the sticks. I was answering to this. I agree for the Sandybridge part. But him he has a Ivybridge. So 16 GB stick will not work on Ivybridge ? I seriously doubt this. Ivybridge can support up to 32gb for any i7 model but the number of sticks is independent (you can have 4 * 8 or 2*16). What is important is the stick memory density, but I am not even sure this is still a matter (was a pretty big matter for ddr2, for ddr3 not sure).

How can you run without no fan with a 2520m ? I have this same CPU and if I am iddling (or watching some film, reading a pdf) it will be at 60°C without fan but the least activity I start it goes up to 90°C and stays here (if no fan). The 2560p heatsink is very small and it was obviously not made for any sort of passive cooling (like a desktop heatsink, they can hold the cpu cool with no fan for a more extended amount of time). Also, just download HPFanControl to control the fan, no need for dst or whatever. You will have a background process but it will work and be good.

My experience with 2560p and 2960xm I had a 2960xm for bargain price (with 4x8gb of ddr3) and as I had a spare 2560p i decided to buy it and test the cpu on this laptop. I had in consequence the laptop with 16gb ddr3 and this i7 2960xm with is the top lane of the sandy-bridge series. I was pretty happy before testing the cpu : 1) i still have also a 2570p with a i7 3820qm and the performances are asto-ni-shing. The 3820qm runs almost at max turbo speed and the maximum temp is pretty high (90-95°C) but it doesnt throttle even with 4 core activated and 100% load (but no gpu load). I lost only 200mhz from max turbo but seems this is due to windows because one time I remove 95% of background processes and I lose only 100mhz. so insteat of 3.5ghz for 4 core the 3820 was at 3.4ghz. I never tested it with 4 cores and full gpu load because I don't game or whatever using gpu. also the 2570p was running on 60w charger. sweet ? So I decided to test this 55w cpu on the 2560p. I swapped the old i7 2520m, booting the laptop with no problem, and quickly arrived on windows. here comes the hassle. first, running benchmarks, I was at 14 seconds on throttlestop. this was pretty high as the 3820qm is at 9seconds best. according to benchmarks on the internet the 2960xm should be between 8 and 10 seconds because it is older gen but runs at higher freqs than the 3820qm so the performance should be roughly equivalent. this means that the heatsink was not enough. I had the 60w charger and also I say obviously it was not enought. also compared to the 2520m i7 cpu, the 2960xm was quicker but not as much considering the price of the updrage (not alot but for a such old cpu as this 2960xm normally this old cpu should be everywhere at 20 to 30 bucks). now, for the temperatures. this IS what I mostly hated. even with core parking, the 2560p was constantly at 70°C at idle. it was impossible to hold comfortably. the 2570p/3820qm is at 60°C. I tried almost everything at least without mooding. I had to keep the fans at higher speed to keep the laptop somewhat cool (= noise). Conclusion From a mooding perspective, it is funny to have a 2960xm is a such small laptop as the 2560p. this is something never hp (even hp) or any other manufacturer would consider. the smallest laptop with a 2960xm i saw was I think a clevo at 13" but this was a custom build and i am not even really sure it was a 2960xm, i just can roughly remember (please confirm). but the very high temps, low performance and noise makes this upgrade somewhat complicated. also, need to buy a much more powerful charger (maybe 80w will be good, my experience showed that the 2560p/2570p don't consume a lot more than 20w without cpu). also some fan mooding or some under coller pad to put under the laptop, unfortunately I don't have this but this would male the laptop possible to hold for extended time and gain 5°C (i think). Hope someone can confirm my experience, even if I agree that 1/ getting a 2960xm is not easy and 2/ buying this for a such old laptop just for testing is complicated, but hey, it is just mood/hack/geek testing ! Edit : it seems that for the 2570p one reason it is so cool with the quadcore was the thermal paste. I used a very advanced thermal paste (silver coumpound) but for the 2560p + 2960xm I used a cheap thermal paste. I will try to repaste with a good thermal paste, I should gain easily 10°C so idling at 60°C instead of 70°C by my rought estimations with the 2570p and another Dell laptop. My rought estimations tells me also that in load the 2560p will still not be able to make the XM CPU run a full spead because of the AC adapter, size of heatsink and HP's bios 40 watts limitation.

What is troubling me is that you have soldered it, according to the pictures, on low heat conductive materials as the surface of the fan and the metal at the surface of the heatsink. I doubt it will be able to remove more than 5°C, even 2 to 3°C. To have better results, you could do the same steps but instead from under the heatsink, I don't know how you could manage this but it will be far better.

Sorry dude, not all of us use their laptop to play video games and watch only YouTube videos. I don't know who you where referring to about the browser tabs but for my case, 10 years ago when I was using Opera 12 to process Internet on a Core 2 Duo ULV (of dual core 1.1 Ghz), I could have at peak 50 tabs open and use only 1.5 Gb of RAM (100% no lie) with an almost not noticeable slowdown. This is incredible in 2018, and most web-page don't do so much more than 10 years ago. There is no 3D animation or some sort of revolutionary advancement that explains why we use 6 times more today. So I don't know what you mean by "basics" but my computer basics where already to do this. I need to open at least 20-25 tabs very frequently in my work to process information on web-pages so this is not just for fun or just the faint of heart. This could be done with a script and command line but the browser engines are far less buggy in processing web-page because of the big competition. A recent video game is consuming less memory than opening web-page on browsers, this would be something of insane 10 years ago. And I don't know if you live on Earth but Windows 10 64-bits is using more memory than 32-bits, and 32-bits PAE (equiv to 36-bits) can do the same as 64-bits for 99% of users. so I don't get your point, and this is not the subject as even on 64-bit users complain. The point is Windows 10 memory management is bad.

Yes... But it is in tech words a very very old chipset. Even on Windows 10 we will not have so much updates, not to talk about Intel policy to release one driver every decade even for recent hardware (so almost never). It is not like you are installing 7/8.1 on a new CPU. I tend to not to care a lot about security on my laptop, if you are not a president/army general/CEO, as most of normal people run on the latest OS and updates, you are kind of the 5% running the "outdated software", and hackers principaly focus on the masses especially if you are not an important figure (why would they spend time develloping virus targetting your outdated software on your outdated computer ? some dude like this probably have only 1$ on his bank account, which of course is not neccesarilly true hah but better focus on the one who spends 2000$ on the latest hardware and software). I was running XP until december 2017, and NEVER had any incident (like some dude robbing my passwords/bank accounts etc) or trojan popups etc. The only reason I stopped XP is because of the uncorrected software bugs in different softwares, for example Chrome 54 crash or even doesn't support latest SSL, more and more websites where getting broken everyday. There was also a whole set of bugs when plugging the computer on external screen, or some random blue screens due to memory overflow of certain softwares.

Exactly, since the Chrome trends memory of apps and browsers increased dramatically, I feel like developers don't program good anymore and they push us to upgrade is making their software heavier and heavier. As I said, I could do exactly the same things as on Windows XP as on Windows 10, the only thing that forced me to "upgrade" but it is more like a downgrade is the lack of support for new hardware/software. Maybe I will go on the Linux side I really don't appreciate how Microsoft poorly made Windows 10 (ugly ass default theme, huge memory consumption, etc)

A picture says tousand of words : I am actually at 10.4 Gb And I have not even open the audio editing software and technical PDFs and DOCX files where it would have easy gone up to 12-14 Gb. (I am using 32bits windows for professional purposes, which is limited to 2.9Gb so I have made a RAMdisk with PAE to access all the RAM) I really don't fall for the standard excuse of Windows 10 using RAM as cache, experience showed me this is not the case (or it is very very very badly implemented) I would like to run 32gb too but just seen the price, 200$

Me too, very quite interested for the same problem. I have made a PageFile of 16GB on my SSD to counter the very bad RAM management of Windows 10 (installed RAM + 16GB PageFile). 32GB of full RAM would lower the SSD wearing (even this seems more like an urban legend than anything else) but the gain in fluidity would be here cause SSD have lower IOPS than direct RAM. On Windows XP, I was using 1.5GB of RAM for exactly the same tasks (browsing Internet, Word document writing, processing heavy audio files and loading PDF ebooks) - and just don't tell me WXP is a very old OS, I just understand why Windows 10 consumes 9.7GB of RAM for exactly the same and the exact same task. Edit: Just watch the price for 32GB of DDR3 for laptops, cost ranging from 200 to 400E, so I guess why not a lot of people haven't tested it.

I did new tests. Seems that I am maximizing at 100°C-102°C on full load with the 3820qm. Maybe I should repaste. Good thing is my CPU keeps full four cores at 3.4 Ghz (turbo 4c is normally 3.5Ghz), and almost don't throttle even at this high temps. Fan is at 100%. I will try a shim and to repaste when I will have more time. So I think you have good results. For professional reasons I am using a mSATA SSD on my 2570p with an mSATA to SATA converter. I don't really understand your question with your WWAN port (by all means it would be too small to fit this adapter) but in the standard SATA place for HDD on the 2570p this adapter you showed should work.

Actually, I have a 3820qm and I had the same problem about the fan spining non stop. This was getting me insane, until a day I disabled by mistake entirely the fan, and it started to speed only at critical temperature at 90°C. So, now my fan is constantly disabled and running at 100% only when at 85°C. I don't recomment this a lot, but my ears are very calm now as the fan in mostly off. I can change laptop, but not my ears. But I have turbo enabled, so the laptop spikes up quickly at 85°C and throtles sometime, especially when it have to open browsers, compress documents, etc. I really don't recommend a Quad Core in this laptop (or even this laptop short) because of all this problems. About the thermal mod, I am not really impressed. Can you show the curves from cold boot (or standby) with HWMon32 ? Just double click on any line and it will show the curves. Because, as it was said, if my fan runs at 100% I will stabilise at 95°C with CPU usage at 100% (stress test). So I think the mod is usefull only to limit the time you hit hight temps but you will always hit high temps. The problem with the 2570p when I watch my desktop PC with massive heatsink (and the way it is constructed) is the 2570p don't have airflow. Also, the mass of the heatsink is to low. With 7 copper plates of few grams this will not be enough to smoothen the temperature spike, and even if it had enough mass there is no airflow to cool it, except the loud fan. So what I suggest is to dramatically increase the mass with large copper papers (not shims), and even if you don't like, to cut of the case. The problem with copper those papers is that it can bring heat on sensible components like the hard drive, so best to have a small paper. Concerning the cutting of the case, why didn't you put a filter ? They sell air filters all over websites dealing with informatics. I tought about a micro fan but : where to plug it in order to control it ? Even if it is micro it will make noise, and we have to plug it to have airflow. Also, I tought about plugging a flexible heatsink but I don't know where to connect it, knowing the laptop becomes pretty hot to touch when the CPU becomes hot, I think a flexible heatsink will not be possible (the palm rest is in metal, so even at 50°C it will be uncomfortable to touch if there is a flexible heatsink connected to it, and let us not forget all the sensible components). HP didn't put some sort of large copper or metal plate (papper) like other laptops of this size, exacltly because even the 35w CPU would be to hot and it would dissipate high heat everywhere. So they put a small heatsink with a (somewhat) powerful fan to disipate heat. In consequence, adding a small copper plate would be enough to passively dissipate the heat, like on other laptops, but with of course an undercloked CPU, then for high performance, the fan will turn on, but the plate will bring the heat everywhere. The fan must be on so long as the plate is not cool. 40 watts in such a laptop is a real pain. I seriously think that the best convinient way to limit fan noise, and heat is simply to underclock the CPU or to deal with it. I use HWMon32 to underclock the CPU when the temps are high,

Actually I have NO issues controlling the fan speed. Funny but you didn't mention the two software you need, HPFanControl and Notebook Fan Control are all you need to complete your list.

I get 100°C at load on a 3820qm, and the CPU throttles to ~3Ghz on 4 cores (the CPU is of corse 4c/8t) while encoding videos for example. I don't use eGPU and don't see a point for a laptop this old in 2017 (its like mooding a dinosaure), but in changing the HD 4000 powerplan I don't see a lot of difference between 350Mhz and 1.2Ghz from a heat perspective, even tough in benchmark I can have 50% more framerate. But, I do see an IMPORTANT difference in temperature when using CPU core-parking on Linux and Windows 10 (disabling 2 physical cores on the 4) where I go down as low as 3.5W for CPU power-consumption and I lose 10°C to 15°C in idle, idling at 45°C, same performance like a dual core. Talking about eGPU, the main problem is that you could get better performance with a newer laptop or using a mini-ITX desktop PC, the problem with the 2570p is heat, noise, and heaviness. The laptop is pretty heavy. If I buy a laptop in 2017 I would buy a cheap ultra-book and a mini-ITX PC because the 2570p is old, and this feels on the everyday usage. Also, the 3820qm is old, and must do a lot of task on the "CPU part" like encoding videos in h265 or watching 4K videos, which makes it noisy and gets the laptop hot. Unfortunatly I think Intel stopped making socketed CPUs after the i7 4xxx and there was no refresh of the 2570 for the 4xxx series Intel CPU so we are pretty much stuck with the 2570p, maybe for ever. I have a 65W adapter and don't have any problems like random rebooting, or battery discharging while burning the CPU, so I will say the contrary to what people usually recommend, 65W is suffisent even with a quad 45W CPU (I am myself shocked but its mainly because the CPU is limited to 40W so this leaves 25W for the rest of the laptop, which is somewhat suffisent). Last, for the eGPU, again you would be better using the last gen ultrabooks because except for HEAVY gaming, pretty much all task hardware-acceleration wise, are integrated in the ultrabooks like 4K encoding-decoding and whatelse. To resume, take the 2570p but only if it is at a good price with the CPU, I would say 70$ for the laptop and not more than 110$ for the i7 quad, this would make 180$ for the total which is good, but maybe not best when considering the 2570p heats a lot, have bad battery life by today's standards, is damn heavy, can be noisy, bad screen quality, and last but not least don't have software or bios or hardware mods even in 2017 which would change a lot of things (like heavy overclocking or underclocking, changing the screen, removing those shitty mini PCIe limitations, and a lot of things I forget) Its true that the old laptops until the series 7 or 8 Intel i7 CPUs (and i5) where like dual cores for the mid-high end CPUs, but now there are quad i5 CPUs, putting the 2570p performance in competition. The only advantage is price. But you must consider the BIG flaws of the 2570p like heaviness (its damn heavy), and other things I mentioned. Sorry for bad english, I am not native english speaker

I didn't compare to the quadcore, I compared to the old 2620m witch is a dual core at 35W, basically the previous generation 3520m, where I had 74°C on stress and you more than 90°C on stress. Coming back to my quadcore at 103°C, for my part I will try (again) the 2560p heatsink to see if I can lose temps. I strongly doubt this will remove 10 to 20°C just by swaping the heatsinks between 2560p and 2570p, but will try again. If I lose temps, I think you istinnstudio can lose temps also, and go down to about 74°C only by swapping the heatsink. If not, I suspect either the thermal paste but, I also did repaste a lot, and many times and didn't get no big heat difference. If I don't manage to lower the temps, anyway I will order the famous copper shim on Ebay, it will delay the temperature accumulation.

Okay, for a 3520m I find at my turn the temp very high, do you run with 100% fan ? I had a 2620m, which is the older gen, but same laptop, same TDP (35w) and had 74°C average max with 2c/4t full stress test. Also, before I had something like 85-90°C just after I pasted the CPU. Then with pure randomness I converted a 1080p video that took like 1h to convert, using the 2c/4t, in ~15°C room, then the temps lowered incredibly low to 74°C, then it stayed about at this maximum average for months. So when you repaste you must burn the CPU to make the thermal paste more efficient. On Windows to run 100% fan you can use Notebook Fan Control, which is working very good (the 2760p fan profile is the same as 2560p and 2570p) But for my new case with the 3820qm, this is not enough. I will change the thermal paste brand to see. batyanko, I will repaste my CPU with a new thermal paste. I will try your tip as I have a multiple partition hard drive with Windows, Linux, and even Android (haha, by the way working great with the 2570p). I will order a new thermal paste brand and try to get the famous CPU copper shim too on Ebay.

I used this for years on a netbook and it WAS outstanding, but not on Linux... On Windows I had like 100% usage to 0.5% on 1080p 60fps @ 25 Mbps videos. You must use MPC-HC to unleash this card. On Linux, was hard for me to setup, and was not very satisfied with the results. Also, on Windows I had to set the output render to "no copy-back" (old video renderer) because the internal CPU or GPU would process back the image and reduce performance. So I guess on Linux you must do the same thing, but I didn't know how.