ThinkPad X220 (i/T) Owners Thread

I've found another much more visible way of seeing the throttling issue on the i7 x220. There's a program called OCCT, which is a GPU (and CPU) testing program:

Download - OCCT Website english

If I run the GPU test (it's the third tab on the left, labelled "GPU: OCCT") at default settings (1027x768), you can keep hwinfo32 running and can actually watch the gpu clock go from 650mhz to 1300mhz, which generally occurs around the same time the cpu throttles down from 3.2ghz down to 800mhz!

And this throttling happens while the CPU temp is < 50C!

My tip is to arrange the hwinfo32 sensor panel on one side of OCCT and the hwinfo32 summary panel on the other side of the screen - that way you can watch all three (OCCT, GPU clock and temps, CPU frequency) things at once.

I found this rather instructive...

Let me know what you think guys?

Petrov.

 

Hoo boy, this is getting bad. I note that the GPU rarely reaches full clock speed when the CPU is stressed, even under perfectly reasonable temperatures. And when the GPU finally does kick in at 1300 MHz, it seems to trigger an immediate drop in both CPU clock speed (to 800 MHz) and GPU clock speed (to 650 MHz). This is most certainly not good and if it can't be fixed via BIOS update means I am going to have to RMA this unit and go with something else. This is the entire reason I paid the premium for the i7 in the first place!

 

So is this really a Lenovo specific issue? It would be nice to see what happens with some of the other brand laptops (HP/Dell) with similar CPU/GPU configs.

 

All,

I am horribly displeased to report that yes, I just now experienced precisely the same situation that petrov reports about the CPU permanently downclocking until a cold reboot. Temperature clearly has nothing to do with this; that is, unless somehow ALL of the temp sensors in the system are incorrect.

This still occurs following a manual driver update of the Intel HD Graphics to 2361.

This is very disturbing. I certainly hope a solution is possible, as if it is not, there is no way I can keep this system (and I'm certain my two customers who just purchased will also wish to return theirs, as they are both gamers). I would strongly advise anyone with an i7 x220 to put their system through the same (simple) testing using OCCT and HWiNFO32, and anyone considering purchasing an i7 model (or who already has), keep a close eye on this situation. You are not getting anything you are paying for unless this is rectified.

 

If they're both gamers, why are they purchasing this laptop in the first place, where there's no option to get a nv or amd gpu?

i think i'm just going to keep the i5 system i received yesterday. this is ridiculous

 

Like me, they didn't buy a laptop primarily to game, but they most certainly will wish to have that option available to them. Both are power users.

The problem here is that we ought to be getting performance which is exponentially better than we're seeing in many games and in other GPU-intensive scenarios. And worse yet, should your CPU devolve into the aforementioned permanently-downclocked state, you must actually power off your entire system and reboot from the off state to correct the problem and reap clock speeds exceeding the extremely slow 800 MHz.

 

The throttling doesn't happen on my i5-2520 machine after running OCCT for 2 minutes. GPU goes to 1300MHz, CPU stays about 2500ish MHz. When I have time to let it run for longer, I will, and I'll report back.

 

since I don't have my X220 now I can't check, but I ran it on my desktop and it doesn't have the correct entries in the dropdown to select the GPU temps...

 

Wow guys, sorry for doubting.

First, I carefully took apart my x220 and replaced the thermal compound. Wow, what a cruddy job they did, as there was WAYYYYY too much thermal compound on it. I replaced it with Shin Etsu G751. Temps went down 6-10c (!) on full load! Amazing. NOTE: if you do decide to do this, you will need a special small socket driver in order to remove the two nuts around the VGA adapter. Those VGA nuts help hold the motherboard in the chassis. NOTE: FAN NOISE HAS BEEN REDUCED OVERALL.

TWO: I am now able to confirm the throttling issue. I have the i7. The easiest way to reproduce the issue is to start by running furmark. In hwinfo32, your gpu will clock to 1300mhz. Then, start intelburntest. Your cpu will now drop to 800mhz, and your gpu clock to 650.

At this point, your cpu and gpu will be stuck. Easy way to fix it is to go into lenovo power manager (Advanced), and switch the System Performance to Balanced, then back to whatever it was before (mine is on maximum turbo).

My cpu and gpu are now throttling fine again.

I will be posting in the thread in the official lenovo forums.

 

So you can break the cycle of permanent throttling by switching the power plan? Perhaps this is indeed evidence that this is fixable via a software-based update.

 

x220t here. When I try to run OCCT, I get an error complaining that DX9 isn't installed/up to date..? Whats up with that? dxdiag shows that I have DX11 installed... GPU test is disabled. :S

 

How high are all your GPU clock speeds getting up to in Hwinfo? Mine never goes beyond ~650mhz, but the meter in Power Manager basic mode shows it's using Turbo Boost and at 1300mhz. i5-2520 here.

 

It does appear so. I will be testing this issue over the coming days.

 

Mine does say 1300mhz in hwinfo32. I'm also using the latest version that was just released a few days ago.

 

Having DX11 installed, does not mean you have the .dll's required for DX9. Believe it or not, you'll need to install the DX9 runtime dll's.

Here: Download details: DirectX End-User Runtime Web Installer

 

I'm using the latest version. So now what. Do I trust hwinfo and believe that my GPU is never using Turbo Boost, or go with Power Manager and assume everything's ok?

 

Yea. Did that. Was sort of confused at first.. because I assumed that DX11 superceded DX9 entirely, so all of DX9's functions would be contained within the DX11 software set. I was wrong.

Just ran OCCT as well. Having your system clocked to 800 SUCKS. (I'm trying to write this and its lagging).

Ran for about 3-4 min and it clocked down to 800. Max core temps were 72/75 for both cores according to HWiNFO32.

Toggling power profiles doesn't seem to unlock it from 800. (Which toggle were you guys talking about? Toggling using Lenovo's power management utility? Which is what i did. Or toggle using windows' powermanagement?)

tl;dr: 800Mhz locking issue confirmed on an i7-2620 in an x220t (x220t has a different mainboard than the x220 I believe, so does that rule out the mainboard as the problem?)

@Vineuro: I saw GPU clock up to 1300 as well.

edit: Also confirmed. Warm reboot does not fix the throttling.
edit2: Cold boot does seem to fix it. Wow that 800Mhz cap is painful @.@

 

@Petrov: Could this throttling be artificially induced? By some weird calls to the GPU? I've run SC2 on this for over an hour with no problems. Maybe its a combination of having HWiNFO32 + benchmarks running that triggers this throttling..?

 

I'm also getting throttling down to 800Mhz more than 4 minutes of OCCT on my i7 2620m despite temps under 76c.

 

So I've reproduced this throttling problem twice now.

The first time, I just went and turned on the Lenovo Turbo+ option in the power manager BEFORE the test.
After running OCCT for about 2 minutes, I got throttled down to 800/650 until I performed a cold boot.

The second time, I did NOT have Turbo+ on. I ran OCCT for around 5 minutes with no problems. Several times I saw my CPU and GPU clock down to 800/650, but when I clicked around (shifted window focus) it immediately jumped back up to 3200/1300. I repeated this several times to make sure that that throttling did not stick. During this test, I went and opened the power options panel and clicked the Turbo+ icon. Right after doing so, I got perma-clocked down to 800/650.


Third try: OCCT has been running for 6 minutes now. Occasional throttling back down to 800/650 but after mouse movement, etc it immediately jumps back up to 3200ish/1300. Have not enabled turbo+ yet. Will do so after the 10 min mark.

Hm. I had OCCT running without the temperature sensors for HWiNFO32 up for the first 6 minutes. After enabling the temperature sensors, the throttling seemed to occur more frequently, until it perma-throttled itself at the 9 minute mark. Correlation..?

 

just wanted to confirm the throttling issue on my x220 i7 system as well...the system is basically stock, received it yesterday and have only installed windows update and google chrome...I would also like to thank those (especially petrov) who have done so much work to try and get to the bottom of this issue...

 

Running for 15 minutes straight. A bit of throttling, but nothing more than 1 sec. Still running at 3200/1300.

Could this be a power issue? In Lenovo's power manager, it shows that the system is currently drawing around 50 watts running this test. With a 65watt adapter (which may or may not be designed to supply a full 65 watts continuously) maybe the system downclocks because of this? Wasn't that what the whole "my system clocks down to 800mhz when I remove my battery" issue all about? Could this be a similar manifestation of that.. except now we're drawing more than what the battery and adapter can supply..?



Running on 25 minutes now. Every time I open the HWiNFO32 Sensors panel, within 5-10 seconds it gets throttled to 800 and stays that way until I close the sensors pane. Immediately after closing the Sensors panel it clocks back up to 3200 (as in within ~1 second EVERY time I've done it).

 

Nevermind, I was looking at GPU memory timing instead of GPU Clock.

i5-2520 goes to 1300mhz and stays there. Above 80C, the CPU goes down to 3.0ghz from 3.2ghz. Peak temp reached running OCCT is 83C.

 

I also only have the speaker box and not headphone box in Sound/ Playback tab.

Should we have it?

Correcting low audio volume in X220 speakers - Lenovo Community

 

Try using CPU-Z and coretemp instead of hwinfo32?

I'd test it right now if I could, but I'm not in a position to do so.

 

A little light relief from the throttling issue, chaps:

YouTube - ‪Star Trek Meets Monty Python‬&rlm;

‎'tis a silly place ....

 

Nope. Switched to it. Still no throttling. Approaching 40 min runtime now.
Weird because I reproduced this twice earlier.

 

Are you running the GPU test alone? Try adding a CPU stress test at the same time.

 

Ran wPrime. Noticed throttling down to 800 when my CPU temp went over 90 (I'm working in a server room.. and its a little warm in here), but went back up to 3200 after the CPU temperature dropped to 80ish.

Only had time to run this twice though.

 

After reading the recent posts regarding the i7 issues, I returned my X220 i7 today for a refund. I am pretty much a ThinkPad fanboy and wanted the latest/greatest but I should have never sold my last X201 i5. Here's what I found:

The Good: Overall nice Lenovo build, better screen than the X201, USB 3.0, faster performance (when it worked).

The Bad: Same CPU/GPU issues as reported by others here, random freezes and lock-ups (NEVER had this with my X201), 7mm hard drive making drive swaps and upgrades a nightmare, HORRIBLE new buttonless track pad (clunky, inaccurate, cursor jumps randomly, TERRIBLE two-finger scroll like they tried to copy the Mac and failed, feels like something out of an old '80s notebook), not really that much better performance then the X201.

Overall the X220 is not really much better or different than the X201 and is actually a step backwards in my opinion. Lenovo could have really jazzed the X220 up for a quantum upgrade but missed on this one. For the first time in a long while, I am notebookless and don't know what to do. Too bad as I was looking forward to this release. I hope this helps others in making a purchase decision. Thanks.

 

Check this one out: FS: Lenovo ThinkPad X201 Flexview IPS 3yr On-Site Warranty

Hope you'll get back to X201 (with IPS). Typing this on an X201 i5.

 

I think I've made a comment before that this current model is flawed in the sense that its cooling solution isn't very well thought out;
ie corresponding cooling can't keep up with the turbo.

OTOH, I'm interested to see how the other manufacturers fare?
I have a HP laptop at work, and its fans run loud even when it's idling.

For this ultra portable form factor, let's look at comparables.
How about Dell's Alienware m11x r3?

I'm not an engineer so I don't know whether there's a better cooling solution for a notebook this thin. Larger copper base and fins?

I don't know how widespread is the throttling issue; perhaps it's a bad batch of GPU cooler, thermal paste, who knows? I just hope that when I get mine in June, it won't be like Petrov's. hehehe.

 

I really don't think it's related to cooling, spam123.

 

So does that mean that when you had turbo boost disable everything is fine even when you had the hwinfo sensors working? Also, I thought that when turbo boost is off, the cpu speed should be the stock 2700, no? You still getting 3200 with turbo boost off? Dont quite understand this part.

One more thing, this might be a similar problem as it was on the 13" macbook pro while running windows. I posted some links on the lenovo thread, but here they are again. It is stated there that the problem was caused while using the turbo boost option.

Quote from article...

Clock problems on Windows

As with the new MacBook Pro 15.4-inch (MC723D / A) also occurs at the 13.3-inch model under Windows, the throttle problem: If the Core i7 at high load is too hot, it lowers the clock rate up to about 800 MHz per core down. This is normal, but he will not cycle more after cooling high. On Windows, this was only after a reboot the case. Since MacOS X everything works perfectly, the hardware is basically fine. So it is obvious to the appropriate drivers for driving the power-saving features, which include the Turbo Boost function.

Apple MacBook Pro 13.3-inch (MC724D / A) | Movmag.com - Gadget's News, Review and Apps

Quote from 2nd article...

The magazine originally thought that Turbo Boost had been disabled under OS X as well, but then tests by AnandTech showed that it was only disabled under Windows. "We first noticed a problem when the benchmarks finished five full runs and the results popped up on screen: the times taken to complete several of the most intensive tests were rising with each run" says PC Pro. "This would suggest an overheating problem, so we ran a temperature monitor to find out how hot this Sandy Bridge CPU was getting."

In fact, the CPU was reaching around 93°C -- almost 200°F. "93°C is not necessarily too high for a modern CPU, but it is the root cause of the bigger performance problem." PC Pro says they're sure the processor isn't turning off Turbo Boost dynamically, since it didn't work at all during their week of testing no matter what the CPU temperature was. Also, the cheaper model with the i5 processor did use Turbo Boost, as did the i7 model under OS X.

MacBook Pro crashes: Apple disables Turbo Boost under Windows? - TUAW

I'll say these are very similar problems to what's going on with the X220, dont you think?

Thanks for all your testing. I've been reading all your posts on the system not throttling unless hwinfo sensors being on. That's good news after all.

 

Also this from another article that reviews the macbook pro 13. I know it might be different since this is on a mac, but it is pretty much the same you guys are seeing in the X220.

Quote from article...

Throttling / Downsampling

The fastest mobile dual-core processor finds its place in a compact 13" notebook and the fans stay quiet? Sounds like a huge challenge that isn't very well dealt with. Particularly in Windows does the i7 CPU quickly reach its thermal limits. On top of that, the power supply is pushed to the max, which might account for the lower clock rate observed under heavy use.

In Windows 7, we observed the processor performance under various levels of stress using the HWInfo32 tool. The processor temperature quickly climbs to near the 100°C suggested maximum temperature. With extreme stress (Furmark + Prime95) we observed an occasional downsampling of the CPU via the EFI (BIOS equivalent) down to 800 MHz. On top of that, the tool also pointed to a thermal throttling of both processor cores. This happens automatically when the CPU reaches a temperature of over 100°C. This however only occurs during the stress test, which is a good deal more extreme than everyday conditions—under normal use this should come up very rarely to never. Other than the very extreme, we tested a few other scenarios.

After running the highly demanding Battlefield Bad Company 2 and Call of Duty Black Ops for a while, we noticed that the CPU was being downsampled to 800 MHz by the EFI. In Cinebench R11.5 CPU benchmark, no such dropping of the clock rate was observed, the CPU staying at a constant 2.7 GHz.

Running 3DMark05 a second time right after the stress test yielded an 8% lower result than the first run-through on a "cold" system. Since Turbo Boost seems to be deactivated in Windows, this might be attributable to the throttling.

Windows 7 is clearly not the best operating system for the MacBook Pro and therefore its performance in Mac OS X is much more meaningful for the ratings we give it. Considering the much better benchmark results in Mac OS than in Windows, we can say Turbo Boost was running in Mac OS during these benchmarks. Using the MSR tool, we observed downsampling to a much lesser degree in Mac OS X. When running CB 11.5 and StarCraft 2 simultaneously, the clock rates of the two cores dropped slightly to 2.3 and 2.5GHz. Apparently, Mac OS X uses energy and cools the system much more efficiently than the Windows driver, Bootcamp.

Review Apple MacBook Pro 13 Early 2011 (2.7 GHz dual-core, glare-type screen) - Notebookcheck.net Reviews

Later it was an update by Apple that they never disable the turbo boost option...(got this from engadget.com)...

Update 2: We've confirmed with Apple that there are no specific hardware or software limits to block the Turbo Boost function, however we've also discovered, through less direct sources, that the company is providing new low level software tools to diagnose cooling issues with the 2011 batch of laptops. Ergo, the speed limits that PC Pro and Notebook Journal encountered might have been caused by inadequate heat dissipation, which arguably is no less troubling than an Apple-mandated de-Turbo-fication.

 

Guys, you misunderstood me. When I said "soft", I meant in feel. Since when was there a noisy Thinkpad keyboard?

The keyboard's feel is definitely worse than the previous X series. I'm just amazed no one can feel it.

 

Personally I'm not very impressed with the keyboard either. In fact, I think my Dell M1330 has a superior keyboard, and it's a 3-year-old one also.

 

I think someone else with the X220 laptop with the i7 processor, not the tablet, should try to do the same tests without HWinfo32 sensors panel to see if the system gets the perm throttle or not or if it only throttles for a little and then goes back to normal. I doubt it since the tests that have been performed already with the laptop show that after it throttles to 800mhz, it becomes a perm throttle and won't go back to normal unless a shutdown occurs. But who knows, hopefully something good comes out of all this.

Also, the same test that you guys being performing (3DMark06 back to back 4-5 times or OCCT while getting the info from hwinfo32) with turbo boost disabled should be helpful to see if the problem is caused only while turbo boost is enabled.

 

Just got my x220 i7 today and was very curious about the cpu throttling issue everyone is talking about. So I tested with OCCT gpu, monitoring with hwinfo32 and cpuz. I let the test run for about 30 minutes. No down throttling at all. The cpu cores at 3.2 and gpu at 1.3 using max performance. No problems. Is there something I am missing?

 

You might just be the single lucky owner of an x220 if that's the case. I'm quite sure nearly everyone else is having these problems.

What's your BIOS version if you don't mind my asking?

Also, do you have a 9-cell or 6-cell battery? I'm using a 6-cell myself.

 

Steve, Have you perform the test with turbo boost disabled? If not, can you tried it to see what happens.

 

If you're referring to "Lenovo Turbo Boost+" (the fan optimization setting), yep, but it doesn't appear to make any difference on my end...

 

What were your temperatures after the 30 minutes? Did you have the HWinfo32 sensors panel open at the same time while doing this (user chaosphoenix stated that while HWinfo32 sensor panels were not open that there was no throttling and that if he opens it, then immediately it will throttle, thats why I'm asking)?

Screenshot by any chance.

 

Ok, just thought it would, but guess not since you already tried it. Thanks.

 

I've finished testing for the day. Just a summary of what happened.

First test: Had turbo+ on to begin with. Ran OCCT and within 2-3 minutes it immediately got permathrottled.

Second test: Had turbo+ off to begin with. Ran OCCT for around 5 minutes. Turned on turbo+. Within a minute or so I got permathrottled.

Third test: Turbo+ off. Ran OCCT for 10 minutes or so. Turned on Turbo+. Nothing happened. Opened HWiNFO32's sensor panel. Saw throttling. Closed sensor panel. No more throttling. Reopened sensor panel and got throttled again. That time it stuck at 800Mhz.

Fourth test: Turbo+ on. Ran OCCT for 15 minutes or so. Opened and closed sensor panel in HWiNFO32. Each time I opened it, the CPU throttled. When I closed the panel the CPU went back up. Someone suggested also doing a CPU stress test. Ran wPrime alongside OCCT for the last 10 minutes. Also switched temperature monitor from HWiNFO32 to CPU-Z and CoreTemp. Throttling did happen, mainly when CPU temperatures went beyond 85C. After it dropped below 85C it reverted back to 3.2ghz. After letting OCCT run for a total of 1 hour, it finally quit. I also closed wPrime. There was no Perma-throttle at the end of this series of tests.

 

No problem, I was hoping you were right

I also attempted outputting video to an external screen instead of the LCD to save power, but that also made no difference.

This is a frustrating problem for sure.

Also guys, be sure to watch the GPU clock when you run this test. It should be ramping up to 1300 MHz; if it isn't, the performance you are getting is not optimal. The FPS literally doubles for the short few seconds when the GPU is properly overclocked along with the CPU.

I have also been running without the sensor panel open; I still get throttling.

 

All right everyone, a little bit of a breakthrough here:

My throttling only seems to occur on AC POWER. When running on battery, there is VERY LITTLE THROTTLING.

Can anyone else confirm this?

 

Wow, I guess this is a real big issue. The way things look I guess we'll definitely be better off with the X220 i5 instead of this. I don't see how this is getting fixed anytime soon. I'll definitely perform the tests when I get my hands on the i7, but it will most probably go back since it seems that all are having this problem.

 

I wonder if this issue may in fact be related somehow to the erratic touchpad behavior I reported earlier when using the 90W UltraSlim charger. Both appear to be charging/circuitry problems--and I wonder if they will require a hardware replacement of some sort.

Also: please don't overlook my previous post everyone. I believe I may have found some critical information with regard to the source of the problem.

 

My God...

My x220 just arrived. The touchpad is garbage.

This is unusable. How could they release it like this?