Razer Blade 14 (7700HQ/1060) beatdown log

Discussion in 'Razer' started by Mobius 1, Apr 26, 2017.

  1. Nuke33

    Nuke33 Notebook Consultant

    Reputations:
    14
    Messages:
    103
    Likes Received:
    48
    Trophy Points:
    41
    Then that is probably the switch for autonomous Speedshift. On the Kabylake Bios this Option is not accessible by default, but Win10 has access to SST (in this case non autonomous Speedshift).
    You can check with HWinfo64.
    It should show SST in green if it supports Speedshift and tells you if its autonomous or not in its tooltip.
    sst.jpg
     
    Vistar Shook likes this.
  2. Vistar Shook

    Vistar Shook Notebook Deity

    Reputations:
    240
    Messages:
    703
    Likes Received:
    565
    Trophy Points:
    106
    Cool....yes mine looks the same, it is set to autonomous. Won´t make much difference, but I noticed your RAM timings, you can tighten the timings if you like, to something like 15-15-15-35, it will reduce latency a litlle. It probably can go lower to 14-14-14-34, but if it you go too low, you might have to reset the CMOS. I have mine at 12-12-12-32 but it is running at 2133 and not 2400.
     
    Nuke33 likes this.
  3. Nuke33

    Nuke33 Notebook Consultant

    Reputations:
    14
    Messages:
    103
    Likes Received:
    48
    Trophy Points:
    41
    The difference is not day and night, but it is noticeable on battery. At least on Kabylake.

    I am hesistant to fiddle with Ram timings on a laptop. Stability is my first priority and in my experience overclocking RAM on desktop boards, it is very hard to find 100% stable timings (especially subtimings). Laptops have too much temperature variation to really guarantee stability at all situations in my opinion.
    If your timings are indeed stable all the time you probably have a very nice latency reduction.:vbthumbsup:

    I only managed to get 2400mhz 10-10-10-30 on my desktop board stable. But it is DDR3 and a desktop board, therefore so much easier to stabilise.
     
    Vistar Shook likes this.
  4. Nuke33

    Nuke33 Notebook Consultant

    Reputations:
    14
    Messages:
    103
    Likes Received:
    48
    Trophy Points:
    41
    @Vistar Shook
    You got me curious as to how low with full stability it can go. I reduced CL/tRCD/tRP timings to 15-15-15 but increased tRAS to 45 cycles, as the Jedec spec would have been. I found it kind of odd razer set timings to 17-17-17-39, since tRAS has to be the sum of CL/tRCD/tRP.

    Tested in a temperature controlled environment at work, it is stable up to 40°C ambient temperature. I didn´t want to go higher with the temperature because of lifespan reduction.

    I will test even lower latencies when I got the time to spare. But right now I am positively suprised. Thanks for the suggestion Vistar ;)
     
    Vasudev and Vistar Shook like this.
  5. Vistar Shook

    Vistar Shook Notebook Deity

    Reputations:
    240
    Messages:
    703
    Likes Received:
    565
    Trophy Points:
    106
    Also try the stability with the command rate at 1T instead of 2T. I am pretty sure that the formula of tRAS is the sum of the others is too conservative, frequently the tRAS is lower than that.
     
    Last edited: Sep 18, 2017
    Nuke33 likes this.
  6. Nuke33

    Nuke33 Notebook Consultant

    Reputations:
    14
    Messages:
    103
    Likes Received:
    48
    Trophy Points:
    41
    I will give it a try, but I think DDR4 is not optimized for 1T and low latencies in general though. It is more designed towards throughput from high clockrates and low voltages.

    I also have seen many RAM sticks not following the Jedec recommendation. Maybe they compensate the tighter tRAS with looser subtimings.
    But my knowledge about SDRAM is not that advanced to properly fine tune them.
    Unlike CL/tRCD/tRP, tRas does have a neglegible impact on performance. I am unsure if it is worth the hassle.
     
    Vistar Shook likes this.
  7. Vistar Shook

    Vistar Shook Notebook Deity

    Reputations:
    240
    Messages:
    703
    Likes Received:
    565
    Trophy Points:
    106
    Certainly adjusting the RAM timings won´t make any noticeable difference, it is mostly for those that can´t stop tinkering...hehe.
     
    Nuke33 likes this.
  8. Vistar Shook

    Vistar Shook Notebook Deity

    Reputations:
    240
    Messages:
    703
    Likes Received:
    565
    Trophy Points:
    106
    @Mobius1, my conductonaut order arrived today and I want to use it on the cpu. On my last repaste I used k5 pro to substitute the thermal pads, but I do have some fujipoly pads, is it worth it to clean up the k5 pro and use the fujipoly pads or the difference is neglible?
     
  9. Mobius 1

    Mobius 1 thwink

    Reputations:
    2,840
    Messages:
    7,928
    Likes Received:
    5,167
    Trophy Points:
    681
    For this system I would use fujipoly if possible, but not necessary.

    Only on the VRAM, but the VRM area works better with K5.

    Also remove the thermal pad as @pbhenry3 said.
     
    Nuke33, Vasudev and Vistar Shook like this.
  10. Nuke33

    Nuke33 Notebook Consultant

    Reputations:
    14
    Messages:
    103
    Likes Received:
    48
    Trophy Points:
    41
    True :D

    Ram timings can be important if it complenents other system latencies. For example if one can´t adjust the timing skews for specific portions of the systembus.
    Sometimes it is even beneficial to raise some latencies to achieve an overall lower total latency.
    I think total system latency is a very underestimated factor by many.
    My dayjob is developing embedded appliances and many of our customers didnt´t even know how much of an impact latency has on useability until they experienced the differences for themselves.
    Interestingly many favored the lower latency even though it also had lower performance.
    They probably wouldn´t have decided so if they had only seen spreadsheets with performance/latency numbers.
     
    Last edited: Sep 19, 2017
    Vistar Shook and Vasudev like this.
Loading...

Share This Page