P50s Owners?

I'm shopping Thinkpads and am between the P50s and the P50. Leaning toward the P50s primarily for battery life with a secondary consideration for weight, and I don't really need the extra oomph or expense of the P50 for what it's going to be used for.

Just before last weekend I had the chance to briefly use a P50s. At the end of my short time with it I thought to check how well it managed thermals/how loud the fans got/how hot the chassis would get. So I loaded up Prime95 while monitoring everything with HWiNFO. The 6500u in the machine would hit around 2.9ghz for all about 5-10 seconds before dropping. Being a 15w chip this was actually expected to me, but I figured it would drop down to it's base clock of 2.5ghz and stay there. Instead I witnessed it plummet straight down to 1.7ghz; and there it stayed. The max recorded temperature was only 58 degrees.

Is this a fluke of some sort? I tried changing any power options that would affect it. I tried it while running off battery and off the wall. Nothing I did changed the results.

Sadly I don't have access to it anymore to try any suggestions anyone might have, but would anyone have any ideas as to why I was witnessing this behavior? Does anyone here have a P50s that can hold me and tell me it's fine and I won't encounter this issue if I buy one?

 

Most low-voltage processors throttle down to their minimum speed when stressed. Since the 6500u stabilized at 1700 Mhz as opposed to its minimum of 800 Mhz, that's actually a pretty good showing compared to others out there.

Bottom line: if you want sustained performance, get a system with a standard-voltage processor.

 

Thanks for responding

My understanding was the u series chips were simply lower clocks, less cores, and wouldn't be able to sustain their turbo clock speeds. Am I wrong? I guess I'm having trouble understanding how intel could market a processor for 2.5ghz at 15w tdp but actually need more than 15w to perform at 2.5ghz. Notebookcheck also doesn't make any indication of less than 2.5ghz clock speeds in their review, unless there was a mistake or misunderstanding on my part. They apparently ran Prime95 on it for an hour as well, whereas I didn't do more than several minutes and it only took seconds to witness the speed plummet.

"The processor cannot utilize its full performance during our stress test (Prime95 and FurMark for at least one hour), because it will only run at 2.5-2.6 GHz after 30 seconds (up to 76 °C). Technically, this is not throttling, but it is pretty disappointing that a ULV CPU cannot utilize its full performance potential in a mobile workstation."

 

I read an excellent article on Anandtech a couple years ago that compared throttling on ULV processors to standard voltage CPUs, and the difference was drastic.

Intel marketing can do all sorts of things to make their products sound good. The difference between Notebook Check's sample dropping to 2.5 Ghz after 30 seconds and yours dropping to 1.7 Ghz after 10 seconds may simply be their sample was a better chip. Binning attempts to match chip quality, but there are still some variations between individual processors. What the graphics portion of the chip is doing can also have an effect.

Intel's current processors are still better than AMD's in virtually every way at all voltage levels. But if you're after performance consistency, ULV processors are probably going to let you down.

 

Would you know what article that was? I honestly can't find anything to truly support this; and it seems inaccurate even if you compare bench scores between the 6600u and the 6700hq. The 6600u scores almost identically in single threaded tests and basically half the score of in multi-threaded tests. If it was truly prone to just being power limited by nature then I would imagine a difference would be visible. I can find laptops that use the lower tdp rating for a given processor thus limiting it, and I can find laptops that have trouble with heat and thus end up throttling. I can also find on notebookcheck laptops that can't properly cool or provide power to both the gpu and processor when both are stressed at the same time, but I should note that their test of the p50s ran both prime95 and furmark at the same time and didn't exhibit this behavior while I was only stressing the cpu. This is what notebookcheck said about another laptop running a 6500u:

"The SoC drops to 2.4 GHz within seconds in the stress test, which is 100 MHz below the nominal clock. Many laptops in the 15-watt class show a similar behavior, simultaneous maximum load for the graphics card (IGP) and the processor creates too much heat, so the system will throttle. We do not think this is a problem unless the load will also drop when we only stress one of the components, but this is not the case here."

So, to me it clearly sounds like they think this behavior would be an issue. I can understand sample variation, but I can't understand a 30% clock difference not being indicative of either a faulty chip or an issue with the power being provided.

 

So I bought a P50s and at first experienced the same issue I was asking about (this time with a 6600u). Found out it was an option in the Lenovo Settings application called Intelligent Cooling that was causing the problem. Apparently it's supposed to detect whether you're using the computer on a desk or in your lap, and throttle the system when in your lap to keep it more comfortable. I don't know how well it actually works at detecting the work surface (I was testing in my lap every time) but turning it off provided me with the full processing power I was expecting. Running prime95 again turbo lasted a lot longer; but more importantly, after that ran out it only slowed down to it's specified 2.6ghz clock and stayed there.

 

I'm glad you figured out the problem and were able to fix it!