Quote from: Throttling on June 16, 2014, 18:05:30Quote from: Florian Wimmer on June 16, 2014, 12:36:39Quote from: Throttling on June 13, 2014, 10:11:11Hi,
You had some concerns over throttling of the GPU & CPU in your tests. You seeked to prove that there were no throttling issues by running 3DMark06 and comparing cold vs warm runs to see if there was a performance difference, you saw that there was no difference in performance & concluded that throttling wasn't an issue. The problem with this is that 3DMark06 does not stress the GPU to 100% load, because it is a CPU limited benchmark with modern GPU's - I know this because I get low GPU load on my system when running 3DMark06. Therefore your conclusion is not a conclusion, it's inconclusive in fact. You also run Furmark & Prim95 at the same time again & unsurprisingly see loads of throttling. Can you stop using Furmark, it's not representative of temperatures and throttling seen in games as the loads are artificially high. The fact that you used 3DMark06 & Furmark to assess if there were any throttling issues with this laptop are quite possibly the worst choices you could have made to provide any clarity as to how this laptop would behave in gaming scenarios.
There are better options out there for assessing the throttling & temperature behaviour of laptops. Please can you use extended (30min) Unigine Heaven Benchmark runs to assess if there are GPU throttling issues. This is very representative of gaming loads, and is also very controllable for you as it runs on a continuous loop. You've used Heaven Benchmark before, but you neglected to use it in this review.
Your site's reviews are really good, but I just don't understand why you hang on to using Furmark, and also 3DMark06. Both tests show nothing about how modern systems perform, they are antiquated. I know you probably find it hard to remove these from your testing suites because you want to have some standards of comparison from one year to the next, but this software has become irrelevant now & actually now provides misleading information that detracts from the rest of your excellent reviews. Other sites update their testing regimes to keep them relevant, I think you should consider doing the same.
we totally understand your concerns. We chose FurMark and Prime95 exactly for that reason: To put the system under absolute full pressure and to see how it performs there. We want to see if the graphics card does throttle at all and if it doesn't do that under full pressure you can be sure that you will have a system that is future proof and can also now run the most demanding software without slowing down. We chose 3DMark06 also for the reason that you mention: It doesn't challenge the graphics card to it's limits any longer. But that's exactly what we wanted: We want to see how the graphics card is performing under medium stress after two hours under full pressure. If there is a difference to the results from cold start (which we have seen often enough with other devices) then we can say whether the graphics card will throttle under normal circumstances when under high but not full pressure or not.
However, your comment is valuable to us and it will certainly influence our ongoing internal discussion about how we review.
Thanks again and best regards
Florian
Hi Florian, thanks for taking the time to respond to my comment, and I'm glad that you've taken it on board to potentially influence the future choices your site makes on future stress testing software. I understand your response & what you have written, and I appreciate that, but I do think your reasoning is still flawed for all of the reasons I mentioned in my initial post. It's OK, you don't need to respond to this one, I'm just pleased that you're willing to take on board comments & suggestions & hopefully have them influence your future decisions on choosing relevant stress testing software. I feel that this is the only place where your reviews are consistently lacking in relevance, everything else is great - keep up the good work, your site is the best on the net for laptop reviews in my opinion, work to keep it that way. :-)
Quote from: Florian Wimmer on June 16, 2014, 12:36:39Quote from: Throttling on June 13, 2014, 10:11:11Hi,
You had some concerns over throttling of the GPU & CPU in your tests. You seeked to prove that there were no throttling issues by running 3DMark06 and comparing cold vs warm runs to see if there was a performance difference, you saw that there was no difference in performance & concluded that throttling wasn't an issue. The problem with this is that 3DMark06 does not stress the GPU to 100% load, because it is a CPU limited benchmark with modern GPU's - I know this because I get low GPU load on my system when running 3DMark06. Therefore your conclusion is not a conclusion, it's inconclusive in fact. You also run Furmark & Prim95 at the same time again & unsurprisingly see loads of throttling. Can you stop using Furmark, it's not representative of temperatures and throttling seen in games as the loads are artificially high. The fact that you used 3DMark06 & Furmark to assess if there were any throttling issues with this laptop are quite possibly the worst choices you could have made to provide any clarity as to how this laptop would behave in gaming scenarios.
There are better options out there for assessing the throttling & temperature behaviour of laptops. Please can you use extended (30min) Unigine Heaven Benchmark runs to assess if there are GPU throttling issues. This is very representative of gaming loads, and is also very controllable for you as it runs on a continuous loop. You've used Heaven Benchmark before, but you neglected to use it in this review.
Your site's reviews are really good, but I just don't understand why you hang on to using Furmark, and also 3DMark06. Both tests show nothing about how modern systems perform, they are antiquated. I know you probably find it hard to remove these from your testing suites because you want to have some standards of comparison from one year to the next, but this software has become irrelevant now & actually now provides misleading information that detracts from the rest of your excellent reviews. Other sites update their testing regimes to keep them relevant, I think you should consider doing the same.
we totally understand your concerns. We chose FurMark and Prime95 exactly for that reason: To put the system under absolute full pressure and to see how it performs there. We want to see if the graphics card does throttle at all and if it doesn't do that under full pressure you can be sure that you will have a system that is future proof and can also now run the most demanding software without slowing down. We chose 3DMark06 also for the reason that you mention: It doesn't challenge the graphics card to it's limits any longer. But that's exactly what we wanted: We want to see how the graphics card is performing under medium stress after two hours under full pressure. If there is a difference to the results from cold start (which we have seen often enough with other devices) then we can say whether the graphics card will throttle under normal circumstances when under high but not full pressure or not.
However, your comment is valuable to us and it will certainly influence our ongoing internal discussion about how we review.
Thanks again and best regards
Florian
Quote from: Throttling on June 13, 2014, 10:11:11Hi,
You had some concerns over throttling of the GPU & CPU in your tests. You seeked to prove that there were no throttling issues by running 3DMark06 and comparing cold vs warm runs to see if there was a performance difference, you saw that there was no difference in performance & concluded that throttling wasn't an issue. The problem with this is that 3DMark06 does not stress the GPU to 100% load, because it is a CPU limited benchmark with modern GPU's - I know this because I get low GPU load on my system when running 3DMark06. Therefore your conclusion is not a conclusion, it's inconclusive in fact. You also run Furmark & Prim95 at the same time again & unsurprisingly see loads of throttling. Can you stop using Furmark, it's not representative of temperatures and throttling seen in games as the loads are artificially high. The fact that you used 3DMark06 & Furmark to assess if there were any throttling issues with this laptop are quite possibly the worst choices you could have made to provide any clarity as to how this laptop would behave in gaming scenarios.
There are better options out there for assessing the throttling & temperature behaviour of laptops. Please can you use extended (30min) Unigine Heaven Benchmark runs to assess if there are GPU throttling issues. This is very representative of gaming loads, and is also very controllable for you as it runs on a continuous loop. You've used Heaven Benchmark before, but you neglected to use it in this review.
Your site's reviews are really good, but I just don't understand why you hang on to using Furmark, and also 3DMark06. Both tests show nothing about how modern systems perform, they are antiquated. I know you probably find it hard to remove these from your testing suites because you want to have some standards of comparison from one year to the next, but this software has become irrelevant now & actually now provides misleading information that detracts from the rest of your excellent reviews. Other sites update their testing regimes to keep them relevant, I think you should consider doing the same.