[CPU + GPU Temperatures + Benchmarks] - XPS 15 [9560] Kaby Lake

Because I know folks that ought not do it, at least themselves
There are reports of higher temps after repasting, stripped screw threads, spilt liquid metal etc. Most were likely not reported.
And apparently there are samples of XPS 15 with no direct CPU or GPU thermal throttling out-of-the-box, which may not need it urgently - the same logic as yours in fact.
Step 1 - learn to benchmark and measure temperatures, then undervolt, no screwdriver.
Step 2 - repaste, add simple VRM thermal pads, evaluate again.
Step 3 - mod further, if you're handy and can take the risk. Or get a heavier laptop instead.

 

I agree with that.

Sent from my SM-G920F using Tapatalk

 

Sorry to have caused such a disruption on this thread!

I appreciate everyone is trying to help best they can, sometimes we get caught up thinking that if someone doesn't take our suggestion, the implication is the suggestion is worthless (not the case).

@GoNz0 - you were likely trying to save me time and money with your suggestion, much appreciated.

Unfortunately my circumstances make it less time (and probably money) efficient to attempt this myself than to outsource.

Even for a simple repaste, I would need to purchase:

[Thermal Grizzly] Conductonaut - $32.15
Anti Static Wrist Strap - $11.99
Laptop/phone precision tool set - $23.98
Delivery of above - $14.99

Materials: $83.11

Add an hour or two of research, plus another hour to perform the repaste and test results (and these are hours I'm either taking from my wife and kid or from work).

It's not as simple as just "repaste" for a guy like me - building in my hourly wage for the hours I'd need to spend on this and suddenly it's a couple of hundred to DYI.

I'm going to PM @iunlock to discuss his services further, but if he can't help me I will go ahead with the above, and post my experience and results (especially over the Aussie summer coming up).

 

I guess all Dell PCs are overheating

 

A good summary for those like to optimize the XPS. myself on-hold on step 2 although during Asus ROG realbench show throttle after 10min, however there is no throttle during gaming etc. I take this as risk/performance balance, I'm appreciate @iunlock provide the guideline on further mod, however those must understand risk.

Please give the thread informative with tweaking improvement result.

 

Hah you two are both being immature right now. You're both valuable members of this community and have helped a ton of people. No one is right in this back and forth argument.
Here are my thoughts:
If you want to do anything intensive at all (aka games or a lot of video editing) then you need to seriously mod the XPS via iunlock's process.
If you use your laptop like most users do and you're just internetting (I love making verbs) then a repaste or even stock is likely fine for you.

I am quite happy with iunlock's modding advice on my laptop. It runs cool and I can even use my laptop like I think it should be used! However after saying that I mostly use my XPS for school (boo)...

Also, @iunlock, sell me your aero 15, I have the itch to upgrade (for no real reason) and the Aero 15 is what really has caught my eye...

 

If you want to do anything intensive at all (aka games or a lot of video editing) then you need to seriously mod the XPS via iunlock's process.

Are you sure that mod is working and xps is not throttling?

 

Get [Thermal Grizzly] Kryonaut instead. Not a huge difference in overall performance as compared to Conductonaut liquid metal, but much less critical if some gets spilled onto the motherboard.

 

Thread cleaned, dozens of posts. Probably some nice babies flushed out with the bathwater, but tough luck. Who has time to read, let alone edit, 4 pages of silly kerfuffling?

The members involved are better than this. We know it, and you know it too!

 

The only throttling I get is when I run realbench, then the CPU reaches it's max power limit and throttles down to 2.4GHz, but that's dell's design and not due to thermal issues.

 

With the recent discussion regarding the usefulness and potential risk of implementing iunlock's mod, I thought I would share my experience having repasted, undervolted, and implemented iunlock's mod with minor changes.

First, undervolting appears to have the biggest impact. Undervolting the CPU and GPU gave me a 10C drop for the CPU, GPU, and Ambient temps. Undervolting the GPU via Afterburner is actually just overclocking combined with setting a max clock/voltage combo. Using the fastest clock (1809MHz) that was stable at 1.000V was the best compromise for me for performance and temps.

Second, repasting with Kyronaut although helpful only yielded about a 5C drop for CPU and GPU in my case. I suspect Dell's original pasting wasn't that terrible in my laptop.

Third, iunlock's mod yielded around a 10-20C drop for Ambient temps. I've been experimenting with different heatsink locations and ways to redirect airflow so that's why the range is so large.

Personally I found iunlock's guide to be perfectly detailed and easy to follow, but I can see why some users may be afraid to tackle it. For one, the potential for a heatsink to move and short the motherboard is very real and very scary. However to my knowledge noone has commented that this has ever happened. If you make sure to cover the entire base of the heatsink with thermal tape, make sure to center the heatsink on the component it is attached to, and place enough thermal tape and pads on top of the heatsinks to ensure firm contact with the copper sheet when the case bottom is screwed back on that risk is largely eliminated.

So my advice for those looking to improve the thermals of their XPS 9560 would be to first undervolt the GPU and CPU to the lowest stable setting (for GPU setting the max stable clock at 0.95-1.00V seems adequate). If your temps in real world scenarios (gaming, data crunching, etc.) are hitting 75-80C I would then repaste and repad the VRMs that contact the heatpipe assembly. If your Ambient temps are still hitting 75-80C after those two steps then I would strongly consider doing iunlock's mod, after carefully reading the entirety of his posts and using common sense regarding the risk involved adding large electrically conductive heatsinks to the motherboard.

EDIT: I should add for those unaware, that the Ambient temperature (for the sensor that reaches high temps) is almost entirely governed by GPU usage and GPU voltage. So if you've undervolted and repasted and your ambient temps are still high, but you are wary to try iunlock's mod, try undervolting the GPU a bit more (max clock at maybe 0.85-0.9V) and see if that compromise in performance is suitable for you.

 

Thanks for the tip!

 

Thanks for your input.

Proof in bringing validation for those who can't seem to comprehend the obvious.



::iunlock::

 

Great thread. A couple of points that are not clear.

1. Wrt. the temperature tables (sensor status). The first picture shows unmodded/ stock. The second picture shows after modding, repasting and undervolting. This does not appear to be a true measure of the efficacy of your mod. Do you happen to have a pair of sensor status charts with everything else the same, only difference being your modding?

2. Is there much benefit to be gained from the air flow to the middle area? It is, after all, hot air from the fans. Would a mod that thermally couples the elements in the middle area to the heat pipes themselves work as well? If everything, ie. the CPU, the GPU and the middle area were to be at more or less the same temperature, wouldn't that solve the problem of the middle area getting too hot and triggering throttling?

 

1. The sensor temperature is not the only measure. The actual problem is that in order to keep this sensor temperature under the line, Intel DPTF downthrottles clocks. The aim is to regain the nominal clocks without exceeding the limit on the sensor temperature.

2. That is not hot air from the fans, rather cool air from the fans before it gets to the hot fins.
A mod that thermally couples the elements in the middle area to the heat pipes themselves might work but hasn't been tested but this has been spotted in other laptops. One would need thick additional copper bands or (relatively) thin heatpipes, not sure if enough conductivity can be squeezed in. And it is not clear if the fin cooling surface is sufficient, mind iunlock adds quite a lot of extra "fin surface". But I think this is the logical way of designing better cooling for a new laptop from the start.

 

2. Good point about the cool air. I had missed that. Thanks.

1. Sensor temperature triggers throttling. Throttling results in cooler temperatures, but the temperature is the driver in this control loop. If the temperature does not exceed the threshold, the clocks will remain high (until other factors kick in, like TDP setting for the chip model). So keeping the temperature of all the sensors below their respective thresholds is key. (Note: they may have different thresholds, depending on the zones that they sense).

Has anyone measured the heights of the different elements in the middle area?

 

The distance between the case bottom & the elements will vary. The case bottom flexes and will vary unit to unit. The elements may vary slighly unit to unit by install.

The principal offending elements are the mosfets and chokes.

Thermal pads between these elements and the case bottom is not particularly effective here. The case bottom is a poor heatsink. And it will superheat air being intaked by the fans, crippling your CPU & GPU radiators.

 

I haven't measured, but my approach appears to work well enough. I used two layers of 1.5mm pads to form the encased zone that goes around the copper sheet, then I made sure to attach thermal pads to the top of heatsinks using thermal tape and added extra layers until each was just slightly higher (about 1mm) than the padded border that goes around the copper sheet. I did this because the thermal pads on the heatsinks will compress more because they are smaller and sit on the teeth of the heatsink and the motherboard and case will bend more in the middle where the heatsinks are.

 

Sorry, should have made myself clearer. I meant height of the various components from the motherboard PCB. The chokes are the tallest, but how much is the difference in height between them and the mosfets, etc.? I mean, what would be the desired thickness of material placed on top of the mosfets to make them the same height as the chokes? In my current laptop, the CPU and GPU are different heights, necessitating the use of a thicker thermal pad for the GPU, or the use of a copper shim. (I don't have a 9560 yet - holding out to see what the next refresh brings before I put my money down).

 

I only put heatsinks on the MOSFETs. I just put thermal pads directly on the chokes. So with a few layers of 1-1.5mm thermal pads you should have no trouble equalizing the height between those components.

 

Hello and thank you.

@_sem_ and @pressing did a great job in providing good points to your questions....

1. When I make any mods, I always make it a goal to have a true measure of efficiency, otherwise it wouldn't be worth all the hassle. During the course of trial and error I collect a lot of data on the fly and continue to work in the direction of what actually works. "Function over theory..." A lot of the data I write down on a note pad along the way, therefore, yes I do record the info, but what is shown is the stock vs after the mod (complete). I hope this addresses your curiosity.

2. Yes! There are huge benefits gained from cooling that middle region. The issue is, due to the design and how close the lid sits above the motherboard, there is very little room for much of anything. So unless one uses the laptop with the bottom lid completely off with a laptop fan blowing on it, which is impractical and defeats the purpose for most people, the only other option is to find a physical means (rerouting of the heat) to address the thermal issue and this is done by modding the system to add what it should have had from the get go.

There are other laptops that have the solution in providing a copper bar to go over the vrm's, connected to the fans

For example, Gigabyte has this on their slim gaming laptop....something so easy to do, yet the XPS 15 is missing it....



Therefore, "The XPS Mod," that I've published is mimicking that idea to dissipate the heat.

Also remember, heat transfers towards the coolest point, which in this case are the fins at the end of the heat sink.

Only if Dell would have implemented something like this.... SMH

Yes indeed. The bottom lid is a very poor heat sink.

.

 

Dell rather prefer fix the flaws with firmware + software How Dell cripple the performance

 

Why Dell did not implemented something to cool the VRM Area? They will tell you XPS lineup is not meant for gaming. Their best argument. Also if you will try to complain about their ghosting issues, they will thell you the same. Isnt there like a tiny fan we could buy and point it towards vrm area?

 

Not enough room for a fan. You could try a Peltier cooling device, I suppose, with the added cost, weight and battery drain.
Just do the mod. It works - at the end of the day that's all that matters.

 

I ak

I have a comment for that. Both CPU and GPU share the same heatpipes. What if adding liquid metal to the CPU means htat ti tranfers heat more efficiently, but this also means that more heat goes to the GPU and this affects it.

10X nVIDIA cardsand the AMD Vega series cards are more sensitive to temperature than the CPU. Little temperature rise affect their perfomance more adversaly than the same temp rise would affect the CPU perfomance. I hypothesise that adding liquid metal to GPU and leave a paste similar to the original for CPU would be better. Getting the GPU to 60 C and the CPU to 76 degrees would make the laptop perform better in gaming. Additionally, with the GPU deactivated the cooling mehcanism should ahve no problems handling the repasted CPU on its own for everything else.

I believe that people are getting freaky about a CPU running all the time at 60+ degrees when they shouldn't. They tend to forget that CPUs are able to handle 80+ temperatures easily for almost a decade now. 15 years ago having a CPU at 90 degrees would be unthinkable, now they can handle it, they shouldnt get there ,but still their limit is at 100+.

Also there are people that work in hot countries, like me, with temps running at 40+ during summer that can't have their CPUs running below 60 anyway even in a desktop. I mean I could, but I would have to go out of my way and buy some super expensive cooling system, which I do not care for. Your CPU wont die from being at 60+ for 5 years. My previous laptop an insiron 2006 model is still running neither CPU not GPU are damaged (Intel core 2 duo).

If all you care about is improving gaming perfomance I would do what i suggested above since GPUs temps are more improtant.

ALL the temperatures are in Celcious.

 

@trofi - If you read the posts here, the critical issues most people face on the 9550 & 9560 are:

- XPS will throttle hard when the vrm temperature sensor hits ~76*C

- there is limited power available to the CPU and GPU before power limit throttling kicks in

The CPU and GPU essentially are bounded by the above so most people who have properly repasted and undervolted are not approaching Intel & nVidia max temps in normal use.

The more extreme mods can get even better performance but that is difficult and requires some trial and error...

OK!

 

What about some light games and their throttling? I dont play the newest games on my XPS, i use it only for light old games like Skyrim, WoW ... Do these still throttle? Also why dont we underclock gpu and cpu so we can precent throttling?

 

Older games like Skyrim can still lead to throttling issues, especially if you have any graphics mods installed. It all comes down to how much GPU usage is required. If you are averaging around 60% or more load on the GPU and you haven't implemented any kind of thermal modification you will still throttle in any game.

In my experience overclocking is actually better for managing temps and throttling. For instance, with the 1050, when you overclock with a utility like nVida Inspector or Afterburner, you are actually just raising the clock at a given voltage threshold. The key to lowering temps is lowering voltage not clocks. So if normally the 1050 maxes out at [email protected] and you overclock it to [email protected], you will introduce a bit more heat and power at first, but if you then restrict that overclock to use the max clock at say [email protected] you'll essentially be getting free performance with less heat and less power consumption than if you didn't overclock. Underclocking wil have the opposite effect, running at lower clocks and higher voltage leading to more GPU usage, more power consumption, and higher temps.

 

Hmm did you try this already? I would want to try it but a quick tutorial would be awesome. I only managed to undervolt gpu via msi afterburner ctrl+f, but that's all. I also undervolted CPU with throttlestop and disabled turbo.

 

The first step is to find your max stable overclock for core and memory. However I wouldn't use benchmarking programs to test stability. I found Skyrim to be very unstable at settings that normally would be fine for 10min Furmark or 3DMark tests. Once you find the max for core and memory, I would downclock the the core and memory by 5-10MHz to give some headroom. Then open the curve editor in Afterburner and follow the instructions in this image:



Those changes resulted in about a 15% performance gain (compared to non-OC settings), and a 10C drop in Ambient and a 5C drop in GPU.

 

I think you posted a thumbnail, as I am unable to enlarge this image for better viewing when I click on it. I saved it to my desktop and zoomed in and it is just barely legible however. lol... Thanks for the insight. Is this possible on the 7577 which has very similar hardware ( i7 - 7700HQ & GTX1060 ) to the XPS?

Also, @iunlock do you think your same mod could be worked into a Dell 7577? The hardware layout is very similar, but with the 7577 having a larger chassis and slightly different cooling output / layout... This thread has me itching to at least repaste my CPU / GPU with some quality compounds to get a glimpse while I'm in there... you know, for science. The thermals are supposedly pretty good on our 7577's in the first place, but top-center above the keyboard (similar area as yours) is the hottest spot without a doubt. I would love to get my temps down as low as yours to improve longevity and reduce throttling that comes with having the i7+UHD vs the i5+FHD most people ordered theirs with. I would love to make a video of the process for others in the 7577 forums and/or get your opinions on modding options based on the layout and size constraints which you handled very well here on your XPS. @custom90gt do you have any thoughts seeing as how you have at least owned a 7577 briefly and did this mod to your XPS machine? Thanks for the inspiration guys!

EDIT:

Also, even when my 7577 machine is under heavy load and throttling due to temps, the air it is outputting barely seems hot. I'm not sure if it is due to the velocity of the air (much higher than my old HP laptop with single fan where the heat coming out of the exhaust was noticeable) or the fact that it is dual fan, but it seems like the spent air coming out the exhaust should be a little warmer to indicate efficient thermal transfer through the heat pipes when it is loaded up. Thoughts here as well?

 

That is the full image. My default font is a particularly small pixel font which is why the text may be difficult to read. Basically after finding a stable overclock, open the Afterburner curve editor and just pick a point on the curve at a lower voltage like 1.000V and drag all the points to the right of it down to the same level or lower. I assume this will work with any other modern GPU, but the easiest way to test is to use a voltage monitoring tool like HWinfo.

 

Totally. I understand the process, I've just never had a modern, powerful system that wasn't completely locked down before. My last HP laptop has an i7 3632QM w/ GT630M that could only edit the clock and memory speeds.

 

Sadly I didn't take the whole laptop apart to check on possible cooling mods. It would require the motherboard to be removed from the laptop. A funny thing is that I have a new in box 7577 that I am going to sell if my XPS doesn't move on craigslist. If my XPS sells I'll probably open up the 7577 all the way and at least repaste. In terms of it needing it, it doesn't. The VRM, CPU, and GPU temps were all very good in the 7577.

 

Wait, you got another 7577?? Lol. That's cool though!

I guess my ultimate goal would be to be able to torture stress test it with the harshest synthetics (Furmark, etc) and it not throttle or overheat at all, but I'm not sure if that's realistically possible even with mods. I don't have an IR Flir gun, but I do have a thermocouple I can use with thermal putty that attaches to my multimeter if I decide to go that route for testing.

I'm planning on using TG Conductonaut / GELid Extreme / Fujipoly XR-m thermal pads and at least repasting mine. Thankfully it has square heat spreaders and not tripods, but I will likely lap all the surfaces involved and possibly even add copper shims to increase contact pressure.

Is there a consensus on lapping/sanding heat sinking surfaces vs polishing surfaces smooth for optimal heat sinking? I was under the impression a slightly scuffed surface yielded more surface area for TIMs to fill in and (in the case of liquid metals) hopefully not be as prone to leaking/pumping out. Thoughts?

 

Yeah I have another sitting in a box that I also got for a good deal. I had someone lined up to buy my XPS but they backed out at the last second so now I have two laptops again... Thermocouples will work great, but getting to the other side of the motherboard will be painful, I'm not sure how well you can test anything. Don't run Furmark on a laptop or even desktop, it's not a realistic test and will damage something even with amazing cooling. I say if you can run Realbench then you're good to go.

 

Oops, hopefully the 30-60 minutes I ran Furmark didn't hurt my 7577 then. I could see it throttling and the temps were definitely high, but the machine never shut off and the fans were running full blast the whole time iirc. I didn't intend for it to run that long, but I forgot about it briefly while eating lunch that day.

What about Furmark set to a resolution that only puts ~90ish % GPU load and doesn't keep it pegged at 99/100% load?

 

Furmark is damaging not just to the GPU itself but to power delivery systems. You can do whatever you want, but you won't catch me running it for more than maybe a couple mins.

 

I'm not too worried tbh... I have tortured lesser laptops worse and they are still kicking just fine. I am in the camp of, if a system fails from something like running furnark for an hour, it wasn't designed right in the first place or was the componentry was faulty to begin with.

Moving on, I have been reading through the 7577 service manual and looking for 7567 teardown images (since there definitely aren't any at all for the 7577) and it doesn't really seem that bad to repaste the 7577. Am I correct to assume all modern heatsink contact points are copper, even if inlaid in an aluminum heat sink mass or spreader plate? Asking because of liquid metals and ordering concerns. I want to make sure I order the right stuff to get the job done the first time.

 

I would not assume that. You need to look and ask someone experienced with your model laptop to have a decent level of confidence.

Ideally you want zero thermal coumpound with the two faces meeting perfectly. The TIM will fill in slight imperfections as it is much more conductive than air. So Idon't think a scuffed surface is ideal for TIM.

One challenge of lapping laptop parts is that they are very delicate and can easily bend making things worse. There are a couple of lapping threads here at nbr although I think most are a bit older.

For liquid metal applications there are some excellent recent threads at nbr. I think @Mr. Fox has some new techniques to reduce leaking risk (e.g. some type of nail polish & foam barrier). I think he recently changed his LM of choice due to leaking. He also has some youtube videos with more details. I don't know if a slightly scuffed surface would be better for LM but you absolutely need flat mating surfaces.

 

Thanks for the response!

My idea of lapping is more or less doing a slow, even figure 8 on some fine grit sandpaper, similar to the old school ways of freshening up a 2-stroke jug head. I know not to really gouge or scratch the surfaces with anything too coarse, I am talking more along the lines of dull scotchbrite finished sheen vs dremel polishing compound / mirror finished surfaces.

The nail polish idea you mentioned is clever. I have seen others suggest leaving kapton tape on after the masking process for just-in-case insurance. I also like the foam barrier concept that could easily be implemented with thermal pad material or thermal putty and at worst case scenario, give you the equivalent thermal conductivity of a decent non-conductive grease.

Thanks again for the ideas and suggestions. I am going to start a new Dell forum post for the 7577 that is more technically and scientifically oriented to contain detailed teardown info and hardware mods/upgrades/insight since there isn't much to be found online right now. I have been waiting patiently to get my 7577 back from Dell since I shipped it out on 11/17/17 to have them replace the keyboard with a faulty semi-colon key. (-_-)

 

One trick for the silence: I've added a stack of thermal pads internally to connect the PCH Chip to the aluminum case bottom.

The PCH chip never becomes too hot, it's rated with a TDP of just 2,6W from Intel, but on the XPS 15 it's cooling is completely passive, it has no fins and no forced air cooling so it tends to accumulate that few heat and it's normal operating base temp was around 45°.

Observing temps I understood that it's temp was the culprit triggering the fans to spin at the minimum speed all the time.
On the XPS 15 the fans cool down only the CPU and GPU but even when they were running below 40° the fans never stopped: silly Dell, the fans have no effect on the PCH and they were spinning for nothing!

Connecting the PCH with thermal pads to the case bottom transfers some heat away and now it's normal operating base temp is around 35°.
I'm using Ubuntu as my primary OS, on idle it runs cooler than Windows and padding the PCH is enough to have the fans off most of the time even with light loads.

 

Hello everyone. I've been having some problems with my 9560 for a while. It seems the performance in games and in general I think has dropped quite a bit, from Battlefield to PUBG. I checked the SSD health with CrystalDiskInfo and all seemed fine, but when I ran Time Spy and Firestrike in 3DMark I think the laptop underperformed. I am looking to improve thermals with repasting and thermal pads, but could I maybe have failing hardware, or is it just thermal throttling? Here are the benchmark results: https://www.3dmark.com/3dm/23759034 https://www.3dmark.com/3dm/23759354

 

Run them again and screenshot your Afterburner graphs and it should be obvious if it's throttling or what else is going on.

 

You can read this thread for some popular benchmarks and then see how your system performs. . .then go from there.

 

Are these indicative of thermal throttling?
https://ibb.co/ddfSub
https://ibb.co/b3XJ0G
https://ibb.co/gjP97w

 

Hmmm... that is debatable... Your temps are definitely under MAX allowed values for your CPU/GPU by their manufacturers (Intel i7 = 105* C and I belive NVIDIA GTX ~ 89* C to 95* C), but I am not sure what Dell's safe thermal limits are set at for the CPU/GPU in your laptop/chassis configuration... it looks close to me.

Your CPU barely tapped out at 100% for any amount of time, but the one square section had it in the upper 90's of usage, which is fine. Based on all the graphs it doesn't look like your CPU is throttling or hitting its limits at all, but it does look like your GPU is throttling slightly, sometimes because of temperature limitations, other times because of lack of being able to draw enough power from its power supply (brick) due to other parts of the system being run at their upper limits at that moment in time. The other limitation I am facing here is I don't normally run the benchmark you are performing, so I'm not sure what its test sequence is supposed to look like.

However, moving on, it looks like your GPU throttled a little bit towards the end of the first square section of 100% usage. I assume those 3 phases are meant to be that way in the test. Based on the GPU graphs of the 3 most intense sections of the graph (that look mostly like big square waves) it would seem Dell's safe temperature limit may actually be 79 *C for your hardware which looks to be around 10-15% of a safety overhead they leave in place to protect the hardware from being thrashed.

The GPU temp limit and power limit graphs also indicate throttling for different reasons in those 3 heavy GPU areas we are analyzing here. The GPU voltage and no load limit graphs are more or less the temp limit and power limit graphs combined into one. The GPU voltage limit and no load limit graphs are for the mostpart, inverses of each other.

So I would say based on all of this, you are throttling a certain amount. It is hard to say if it is a normal amount for the benchmark being performed on your same machine however because I am not familiar with that benchmark and its test phases or what it throws at your machine nor your specific machine chassis and hardware configuration

If you are comfortable running furmark, you can stress your machine to its upper limits for a few minutes after it is heated up and get reliable upper limit graphs from Afterburner to compare the benchmark findings to.

I would also say, make sure to maximize your Afterburner window for maximum time span. Also you can set your polling rate to 200ms to 500ms for an even smoother, more accurate graph, but it will scroll across the screen quicker, meaning you will actually have a narrower time window to look back at your system's performance graphs. This is where the Afterburner Remote Server App can sometimes come in handy also.

 

Read the thread man.

That may be throttling relating to vrm temps

 

going to post my benchmarks when i get up and running but i have a issue that's driving me crazy, hopefully someone here has encountered it.

here is a screenshot of how it is now, original score was over 3000
https://imgur.com/a/GhFF2

I did a benchmark on my ssd when I took it out of the box, did a fresh install still getting good speeds. Now it's about half speed. Windows 10 is installed uefi w/ secure boot using gpt. ahci is enabled in bios. i've installed all appropriate dell drivers, and used the intel tool to get the latest sata ahci drivers. ive exhausted any "ssd optimization" type of fixes i've found in different places online. i've noticed in the benchmark that before it listed "iastora" and now it's "stornvme" but regardless of what I do i can't get it back to iastora even installing the intel rapid storage drivers. i've booted into safe mode tried removing and installing the intel drivers again, even tried editing the registry to force ahci.

i'm hoping someone has went through this and can help me out. thanks

 

Hey guys, here's currently what i've done so far:

Repaste with Artic Silver 5, Heatsinks on the MOSFETS, Undervolt CPU -115 and iGPU -100, and recently did copper sheet mod with thermal pads. The only thing i didn't do was apply new pads to the VRAM (stock looks like its making great contact (very tight). I'm still getting the power limit with my GPU..any suggestions? See pictures.

When i repasted with Kryonaut i was getting way hotter temps than I was with the AS5.