Mythlogic Pollux 1613 / Clevo P157SM Review
by Jarred Walton on August 30, 2013 12:00 AM ESTMythlogic Pollux 1613 Overclocking
One of Mythlogic’s claims to fame with their Clevo-based notebooks is that they’ve customized the BIOS to allow overclocking. I haven’t tested any other Clevo P157SM/P177SM notebooks so I don’t know if they’re done a lot or if they just tweaked a few items, but you’re able to raise the maximum CPU multiplier. I’m not sure if there’s a different limit imposed by Intel, but for the i7-4700MQ we could increase the single-core, dual-core, multi-core multiplier by two bins (3.6GHz max) whereas on the i7-4900MQ we’re allowed to go up to four bins above the maximum single-core clock, so 4.2GHz for all settings. It’s worth noting that we were not able to run at 4.2GHz stably, and in fact I couldn’t get the Clevo P157SM to even boot into Windows. I didn’t do any intense research, but ultimately I found that setting the CPU to 4.0GHz for all loads ran without crashing, so that’s what I tested.
Before we get to the charts, there’s on other tidbit I want to discuss that relates to Haswell. With Sandy Bridge and Ivy Bridge – and pretty much every CPU I can recall testing over the past decade or so – I’ve found that best performance is achieved using the High Performance power profile. However, at one point in the testing of the Clevo P157SM I ran some tests on the Balanced power profile. In most cases the difference wasn’t anything noteworthy, and often the High Performance profile was indeed faster, but in some games the choice of Balanced actually improves performance quite a bit – GRID 2 being the main beneficiary. Our best guess is that with some titles, selecting High Performance tries to run all the CPU cores at maximum clocks and may end up lowering the maximum Turbo Boost multiplier. With GRID 2, at our Value settings the Balanced profile ended up 15% faster, and even at our Mainstream settings there was still an 8% increase in frame rates (though this went away at the Enthusiast settings).
I mention this because all of the initial testing with overclocking and stock CPUs was done using the High Performance profile. I could go back and rerun every test on the Balanced profile to investigate the topic more, but that’s a lot of hours of running repetitive benchmarks for a relatively small payoff, and I’ve already benchmarked the P157SM far more than any other notebook in recent history. If the High Performance profile is indeed causing maximum Turbo Boost clocks to drop slightly because of increased heat, however, overclocking may further exacerbate the problem. If you end up buying one of these notebooks, I’d suggest at least looking at how the Balanced vs. High Performance profiles affect the games and applications you use. Of course, for laptops that I actually use, I tend to just stick with the Balanced profile as a good blend of battery life when I’m unplugged and performance when I’m plugged in, but it is an interesting change with Haswell regardless – possibly brought about by the integrated voltage regulator? Anyway, on with the overclocking benchmarks.
The same titles that benefit the most from the CPU upgrade should likewise benefit from overclocking. Skyrim, GRID 2, and StarCraft II apparently love having faster CPUs… but in practice, the gains from overclocking the CPU are small at best. The 4900MQ actually shows what I was talking about with the higher maximum clocks potentially ending up slower on average compared to the stock 4900MQ. In the end, the 4700MQ benefited more from overclocking than the 4900MQ, but perhaps with a bit more tuning I could coax the 4900MQ higher – like limiting the clocks to 4.0/3.9/3.8/3.7GHz for single/dual-triple/quad loads instead of setting them all to 4.0GHz. The short summary is that on average the stock 4900MQ is around 5-7% faster than the stock 4700MQ, while the overclocked 4900MQ is only 1-5% faster (and sometimes slower) than the overclocked 4700MQ.
Overall changes from overclocking the 4700MQ are anywhere from -3% to +8%, with larger changes at Value and Mainstream settings; the average change is +0.7% at Enthusiast, and +2% at Value and Mainstream. Overclocking of the 4900MQ shows changes ranging from -10% to +3%; GRID 2 and StarCraft II both drop in performance by more than 5%, so further tuning would be required to find a good balance. The 2-bin overclocking limit of the 4700MQ also prevents it from fully closing the gap with the stock 4900MQ, but it does get a bit closer.
There’s obviously a second question that I haven’t even touched on with regards to overclocking: can we get any additional performance out of the GTX 780M? Personally, with the GPU already hitting 85C under normal gaming use, I’m not inclined to push it harder. This is an expensive notebook with a very fast GPU, and until we get even higher resolution displays there’s really not much need to overclock the GPU. Others are welcome to do so, and I know some have already done so and posted results, but if I purchased a notebook like this I wouldn’t want to risk damaging a chip for a few more FPS.
Facebook
Twitter
RSS
47 Comments
View All Comments
GTVic - Saturday, August 31, 2013 - link
Obvious answer is the tradeoff with the weight. If you overclock to the extent that you need 220W+ then you are going to need heavier duty cooling design which adds even more weight.Khenglish - Saturday, August 31, 2013 - link
The 15" and 17" laptops both have identical motherboards with Identical cooling (the 15" has a default aluminum CPU radiator, but most resellers offer the copper version found in the 17")So no there would be no extra weight.
waldojim42 - Friday, August 30, 2013 - link
I, for one, am always glad to see high end displays in use. Props to Clevo for such a decent display.On your last page, you mention wanting to test against the Alienware 17. While I understand the reasoning, I think that Alienware/Dell put the majority of their focus on the 14 and 18 this time around. Both the 14 and 18 have high end display options that the 17 doesn't get. IPS, and PLS to be exact. Frankly, I would be far more interested in your results on those over the 17" TN panel.
Also, I have a 9700QM based machine being delivered later this week - one thing I didn't see mentioned, was the thermal impact from overclocking. Also, is the 2bin OC limit an Intel limit, or BIOS limit? I know the machine I am picking up is supposed to allow overclocking, but I prefer not to go into these things blindly - and there really isn't a lot of good information on this.
JarredWalton - Friday, August 30, 2013 - link
I honestly don't know if the limit is BIOS or Intel, but I suspect it might be BIOS. Usually, Intel lets you do +4 bins, and you can set SC, DC, TC, QC to the same multiplier (basically, what I could do with the 4900MQ). But perhaps Intel locked down the 4700MQ for precisely the reason that it would be nearly the same as a 4900MQ if you could set everything to 3.8GHz. Thermally, the 4700MQ didn't have any issues overclocked, but the 4900MQ behaved oddly; I think I need to tweak the OC settings to get more out of it, and even then I think it will only be at best a 3-5% improvement over stock clocks.Finally, regarding Alienware, the 18 is hard for me to go for, as it's so huge -- it's in the same category as the Clevo P370SM. I'm just not sold on SLI for notebooks, as there are too many gotchas (heat, size, drivers, cost, etc.) I didn't realize the AW 17 didn't do anything more for the display; I was hoping to see IPS/AVHA/PLS in there, but I guess not? Maybe next time; anyway, Dustin has the full review in the works, but the high gamut TN panel in the Clevo's is getting somewhat out of date.
Meaker10 - Friday, August 30, 2013 - link
4700 = +2004800 = +400
4900 = +600
4930 = unlocked
4.2ghz on all cores is a +600mhz boost to the core ratio:
4900 = 38/37/36/36 (4core/3core/2core/1core turbo max)
4990 oc max = 44/43/42/42
The biggest restriction on the 4900 and 4800 will be that tdp increases are locked to +10w to 57w so going beyond 3.9ghz wont mean much (setting anything higher than 57w may looks like it has set but the cpu power wont go beyond 57w).
Also powernotebooks is niw selling a 4.1ghz oced 4939mx and 3.9ghz 4900mq in the msi barebone range. How many times do we have to tell you your machine was a bad sample?
kogunniyi - Friday, August 30, 2013 - link
You're right, but MSI deserves some flack for sending Anandtech two bad samples.DanNeely - Friday, August 30, 2013 - link
They've gotten a good one; hopefully at part of one of their upcoming articles they'll publish updated thermals. Unless the issue was a, now fixed, design flaw/fan controller bug though 2 faulty laptops in a row suggests serious QA problems that would scare me away.JarredWalton - Friday, August 30, 2013 - link
Anyone that thinks the MSI was a "bad sample" at this point is fooling themselves. You can keep "telling us we had a bad sample", but that just isn't true. Dustin had one MSI notebook, and at full load it hit 96C on the CPU (unless you trigger the max fan tweak that turns it into a blow dryer). Then he sent that back, and NVIDIA sent me not one but two MSI GT70 notebooks for additional testing. Both of those hit 95C on the CPU under full load -- 10C hotter than the P157SM reviewed here. That said, the MSI notebooks I had from NVIDIA did not throttle -- they just ran hotter than I'd like. 95C is still less than the 100C max that Intel specifies, but how long will you be able to game on a notebook hitting such temperatures before things start going south?Could we replace the TIM to get reduced temperatures? Perhaps, but I just don't think that should be standard procedure, and it shouldn't be required in the first place. What's more, it still won't run as cool as the dual HSF Clevo designs, since it's cooling two power hungry chips with one fan.
As for the overclocking statements above, the maximum clock on the 4900MQ at stock is 3.8GHz, and pushing that to 4.2GHz was not stable (wouldn't even boot into Windows). While 4.4GHz is an option in theory, in practice I don't see it happening on any but the best CPUs. Even 4.0GHz is behaving oddly, with performance often lower than stock settings. I'm going to try a scaled OC of 4.1/4.0/3.9/3.8 to see if I can get any better performance than my initial results though....
Meaker10 - Friday, August 30, 2013 - link
The thing is to go beyond 3.9ghz you will start having to add voltage which will instantly slam you into the 57W TDP limit. It looks like you were already hitting into it anyway.This is an even more delicate operation than a mITX desktop after all but can be done if you are patient.
But by your own admission if throttle was not occurring on the sager then temps would likely be in the same area and the dual fan has no advantage....
JarredWalton - Friday, August 30, 2013 - link
Temperature and power draw are different, so the 57W TDP is the factor. The Sager can better dissipate that (keeping the CPU at <90 and usually <85, with the GPU <80), but Intel's preventing higher Turbo even at lower temperatures. The MSI isn't throttling, meaning, it's not dropping below the rated speed, but it's also not running at max Turbo either. There's a difference.