Tutorial #27: TWEAKING THE ASUS TUF GAMING A15 - PART TROIS

 


Update (June 1): 
The curve I ended with in the original article ended up being unstable in scenarios where the frequency dropped down to the 1600s on account of the thermal limit being breached. Hence, I tweaked the curve further to increase the voltage for the lower frequencies. Turns out it is not straightforward as even minor voltage-frequency changes at the knife edge can result in crashes. Eventually, I settled on the curve below which happens to be taken at 41 degrees Celsius ambient temperature and is thus levelling off at 1800 MHz, though it comes down by 15 MHz at over 50 degrees, making the maximum frequency same as the original curve.

Despite the reduction in undervolting, it actually lead to an increase in overall performance at a lower maximum GPU temperature. Nice!

Finally, the comparison with stock performance. The efficiency is on par with stock as diminishing returns kick-in for the GPU at its limit. Considering that the GPU TGP is now 25W higher, the chassis is doing a good job of handling the additional heat. At the same time, the CPU performance is now up to stock, but at lower power and temperature values. Overall, this is a really nice setting as the performance has increased by nearly 15% from stock and is ever so closer to desktop performance, at a much lesser TGP.


Original Article:

Tinkering with the device can be a progressive exercise. Part 1 was about getting more performance out of the box whereas Part 2 was a more holistic change aimed at efficiency. However, there is always more to be found by pushing the limits and this article is all about it.

In my previous article, I had lamented the 90W TGP limit on the GPU, especially as I wasn't close to hitting the thermal limit on the GPU after the tweak. As it turns out, extending the TGP limit to 115W that this GPU is capable of is quite a trivial task.

Even though I mention trivial, it carries a certain element of risk and hence anyone attempting the same is forewarned about the consequences. Having said so, it is a rather quick and easy process. Firstly, you will need NVFlash to take a backup of the current VGA BIOS and also to flash a new one. Secondly, you will have to find the correct replacement BIOS to flash. Any of the bios files with a Boost Clock of 1560 MHz are the 115W ones but in my case the one that worked was not an Asus one, which resulted in the USB-C DP output not working but the MSI one listed here. Note that you have to use nvflash with the "-6" option to override the PCI Subsystem ID error as follows in the case of the above BIOS file.

nvflash -6 232273.rom

Voila! You now have the full TGP unlocked. At first, I decided to give the official overclocking method a go which can be accessed from within Geforce Experience by enabling "experimental features" in the Settings and then selecting the "Performance" option from the overlay (Alt + Z). The automatic performance tuning option displayed a boost of +75 MHz though in GPU-Z, I could see that all the clocks had been increased by 100 MHz. At the end of the day, it produced a TimeSpy score of 6840 which you will recall is higher than the maximum score from last time. The GPU can be observed to now use the full 115W, though it also hits the thermal limit which was not possible with the TGP at 90W. All things considered, this overclock is on the conservative side.

The better solution then is to use the legacy scanner which is more at the edge of performance besides allowing a more generous memory overclock. One thing I failed to mention previously is that the OC Scanner on MSI Afterburner now uses the official NVAPI method of overclocking which does not detect cards lower than the 3xxx series. Hence, to get the Voltage-Frequency Curve, you will need to use the legacy OC Scanner instead which can be easily enabled in MSI Afterburner by editing the MSIAfterburner.cfg file in the root installation folder ("C:\Program Files (x86)\MSI Afterburner" by default) as follows:

LegacyOCScanner = 1

Doing so immediately produced a better result that breached the 7000-mark with the 500 MHz memory clock that I usually run the card at. It wouldn't surprise you to know then that it hit the GPU thermal limit along with the power and utilisation limits. 


The next step then was to see if I could extract the same performance at a lower temperature and power consumption. As you can imagine, it was quite a time-consuming exercise to find the ideal voltage-frequency curve. I did manage to get the maximum Time Spy score of 7155 with a maximum clock of 1800 MHz at 825 mV but that too hit the thermal limit, besides being unstable. Eventually, I found the sweet spot at the max frequency of 1785 MHz at 818 mV, going any lower resulted in instability. I also experimented with a lower frequency of 1770 MHz at lower voltages but that brought down the overall score, besides surprisingly resulting in a higher peak temperature, albeit at a lower power consumption.


Drum rolls then as I reveal the final results for the above curve which happens to be my (final?) choice for the Turbo fan profile. A score of 7148 with an average CPU and GPU power consumption of around 16W and 61W respectively is quite good in my opinion. At the same time, it did not even hit the thermal limit during the test. All in all, the best profile I could hope for at present for this hardware combination.

However, an important thing to note is that an absolute performance profile does not work well for lower fan profiles. Case in point, when I switched to the Performance mode with this profile, the score dropped to 6581 points which is far too less. Turns out that the higher voltage-frequency combination is far too much for the lower fan RPM. 

Thus, for the performance profile, I ran the OC scanner again and applied the +500 Mhz memory overclock to come up with a score of 7045 which isn't too far from my absolute profile with the Turbo mode. However, it comes with a huge spike in GPU power consumption compared to my Turbo mode settings besides still hitting the thermal limit.


I tried to tweak the curve further for the Performance mode but the lower fan RPM requires the clocks to be dropped significantly. From my tests, this would require dropping the max clock to 1600s at may be mid-700 mV but that also drops the score quite a lot, in the region of 6800-6900. That might be a feasible solution but for now I have left it at the default OC Scanner curve as I wouldn't be using the Performance mode often.

Final words also on the clock speeds reported in GPU-Z. The official Geforce Experience performance tuning with NVAPI displayed higher GPU clocks in GPU-Z, so you may be mistaken in terms of it offering better performance. As you can see below, this is a screenshot of the clock speeds reported for my final profile which doesn't seem that significant but offers significantly more performance. It is all about the frequency-voltage combination rather than the reported figures, so don't simply go by the higher numbers as it doesn't always offer the best performance.


With this, I hope I have finally closed the chapter on tweaking the hardware for the Asus Tuf A15. Before I go, here is the final comparison with the stock settings...


…and also the previous tweak at 90W. As I have mentioned previously, incremental GPU performance comes with a comparatively higher power consumption and hence the efficiency is still highest for the 90W tweak. However, at the same time, the 115W tweak still offers better efficiency than stock and hence it is a win-win in my opinion. Hope you have a great time tweaking your hardware and let me know if you have any queries.

 

Tutorial #26: Debloating Samsung Galaxy (Tab S5e)

 

The Samsung Galaxy Tab S5e is a great everyday tablet as I previously mentioned in my review apart from the fact that it isn't so out of the box. The reason behind this is Samsung's bloated One UI which takes a toll on the hardware, especially one with a mid-range SoC. However, this issue is easily fixed if you are okay with tinkering with your device and voiding the warranty.

This tutorial is a means to that end and lists the process I follow in getting my S5e working as fast as a custom ROM, and I can attest to that having also tried Lineage OS on this device. The good thing is that Samsung has stepped up its game in terms of updates. This doesn't imply you will always get the latest updates for your region, but there is a way around it and that is by flashing the firmware manually which in the case of this device can be tracked over here. However, I would recommend downloading the firmware itself at full speed from Samsung using tools like Frija or Samsung Firmware Downloader.

The other thing to note is that you will need to root your device to access it at the lowest level and I wouldn't describe that process in detail since it is already well-documented. However, what I would like to clarify is that although rooting along with unlocking the bootloader permanently trips Knox on the device and thus renders it useless for any functionality utilising Knox (like Mobile Device Management by corporates), the device itself can still be operated without root and will thus run enterprise apps managed by Microsoft's Intune Portal.

The first thing I do towards speeding up the device is to flash Multidisabler. Hit up the link to read more about it but be aware of what it does. On the whole, it has the largest impact on speed since it disables file-based encryption, besides addressing some hijinks resulting out of rooting the device.

The next important file to flash is the BluetoothLibraryPatcher. Samsung devices stop retaining paired devices when rooted and this comes in real handy, especially if you are using wireless earbuds and fitness trackers. Although you could use the Magisk module, I flash it using TWRP as I don't use Magisk for day-to-day stuff.

Now we come to the meat of changes, moving from generic to specific changes. This is simply about deleting apps that you won't be using at all and hence, by definition, this is a personal list. I have listed the apps which I have observed as being unessential to my need, but your needs may differ and thus will require some experimentation on your part. You could delete these apps using a file manager with root access or create a script to this effect.

These systems apps are essentially distributed across two partitions - system/app/ and system/priv-app/ and may vary depending on the region. In my case, I always use the latest release from any region and thus the list may be cumulative in that regard. As of writing this, it is based on the UK release as it was the first to receive Android 11. The ones highlighted in "cream" indicate new additions to Android 11 whereas the Android 10 apps that are no longer in 11 are highlighted in "red". I have kept them in the list for posterity but will remove them in the near future once I am convinced I don't have any reason to ever go back to Android 10. With that, I present you with THE LIST.



As an end note, I would like to state that you don't need root after the completion of the above process in terms of debloating. You can of course use continue to using root for other apps but that is not an option for me as I run some enterprise apps on my device that don't work with SafetyNet failure. If for some reason, you cannot boot to the un-rooted system after completing this process, then remember that it is possible to do so by booting into the recovery and selecting 'Reboot to system'. This boots the un-rooted system and should you need to boot into root again, you can do using the key combination listed above on the Magisk readme page.

With this, you now have a un-rooted system that is completely debloated, passes the SafetyNet and is just as fast as a custom ROM.

Musing #75: Analysing the Valencia Formula E debacle

 


Some may call it exciting and others farcical. For a fledging series, Formula E certainly attracted the wrong kind of attention with the slipshod finish yesterday resulting in only 9 drivers being classified at the end. While it is easy to call out the FIA or Da Costa for the extra lap, the reality, as always, is more complex. This analysis thus aims to clarify the events as they unfolded.

The first big question that I came across on the web was about the big change in energy reduction percentage during the various SC periods. While there was only a 3% reduction during the 3-minute safety car (SC) at the 20-minute mark, the final 5-minute SC resulted in a 12% reduction. Well, this is quite easy to explain keeping in mind the starting available capacity of 52 kWh. Essentially, the percentage value displayed on the screen is a relative value whereas the absolute reduction is happening in terms of kWh, both to the usable energy as well as available energy.

At 20:38 remaining, we can see that the available energy is 61%.

Moments later, it drops to 58% after the reduction.


The reduction itself is 3 kWh for that SC period and a total of 9 kWh for the race.


This latter part is the most important as it indicates the total available energy after the reduction. Back of the envelope calculations for this scenario is as follows. Note that these calculations are based on the whole number figures displayed in the TV graphics while the actual numbers with the correct precision would be slightly different.

Usable energy before reduction: 46 kWh (52 - 6)
Available energy before reduction: 61% of 46 kWh = ~28 kWh

Usable energy after reduction: 43 kWh (52 - 9)
Available energy after reduction: 58% of 43 kWh = ~25 kWh (28 - 3)

Now let us take a look at the final SC. The available energy is 18% prior to reduction.

It drops to 6% after reduction.

The total energy reduction is 19 kWh in total and 5 kWh for that SC period.

The calculations now are as follows:

Usable energy before reduction: 38 kWh (52 - 14)
Available energy before reduction: 18% of 38 kWh = ~7 kWh

Usable energy after reduction: 33 kWh (52 - 19)
Available energy after reduction: 6% of 33 kWh = ~2 kWh (7 - 5)

As you can see above, the percentage value can be quite confusing for the viewer as both the numerator and the denominator change by the same amount and thus the change in the actual percentage value will be more drastic for lower energy values than the higher ones.

That explains the TV graphics but then why were the teams caught so unawares towards the end? For, let us go to the period just as the final SC came out.


At this point, Da Costa has 22% of 38 kWh usable energy i.e. 8.4 kWh. The fastest lap as can be seen in the office notice was about 1m 40s. It thus indicates that at this point, there was enough time to cover 5 laps.

However, when the SC came out at 5:38 remaining, there was still a part of the lap remaining. Luckily, Da Costa floored the car at about the same place after the SC, so we can easily make out the time needed to reach the finish line from that part of the track.


As can be seen from the images above, it takes about 25 seconds implying that Da Costa would have crossed the line with 5:13 remaining if there was no SC. Considering the fastest lap of the race, 3 laps would have taken around 5 minutes. Thus, if Da Costa was to complete 3 more laps in 1:44 to 1:45 minutes, he would have needed to complete only 4 laps to finish the race with about 8 kWh remaining which is perfectly feasible.

The problem then is that the SC pace was not enough to scrub off a lap and thus the cars still had to complete 4 laps to finish the race, around 2 laps under the SC and a little more than 2 laps at race pace. While the race pace target was 2 kWh/lap, under the SC, assuming a lap time of 2m 30s, the reduction would have been 2.5 kWh/lap. This implies that the cars lost around a kWh of energy behind the safety car due to the time elapsed and at the same time utilised closed to 1.5 kWh of energy following the SC.

Thus, while in race conditions, Da Costa was expected to have 4.4 kWh of energy at the point where he started the final run with 2 laps to go, in reality he had about 2 kWh. The only feasible option was to limit the race to one lap after the SC which he was unable to do so. At the same time, Mercedes seems to have a 5-lap target in mind before the SC and thus were keeping more energy in hand compared to Da Costa who was hoping to limit the race to 4 more laps at the time the final SC came out.

While it was a shambolic end to the race with the FIA shifting blame to Da Costa for not controlling the pace, the FIA is not without blame. They had never made a provision for such a scenario and the fixed reduction of 1 kWh/min applied by the FIA is excessive when the usable energy is low.

One way of tackling such a scenario could have been to reduce the energy allocation for a race lap over a SC lap duration (i.e. ~2 kWh reduction over a 2:30m SC lap). This results in a reduction rate of 0.8 kWh/minute. If this seems too low a reduction, then the rule can be changed to apply this limit only for the final 10 minutes of the race.

The other option could have been to not deduct the energy consumed by the car behind the SC which seems to be about 0.3 kWh/minute (can be calculated from the fact that usable energy reduced from 22% to 18% behind the 5-minute SC with 38 kWh available energy). However, this would result in a reduction rate of 0.7 kWh/minute and thus even more benevolent that the previous approach which is in fact more practical as it takes in to account the energy consumption by the race car and SC for a specific track.

Will this situation be addressed? That is anybody's guess as Formula E certainly seems to be quite disorganised at present. However, the solution to the problem is available as highlighted above and all it needs is the FIA to act on it. Most probably though, I think the teams will adjust the software to account for the SC loss going forward and we might see a slow but secure finish if such a scenario arises in the future.

Musing #74: Designing Fossil Hybrid HR Watch Faces


  

Time pieces move on with time and the Fossil Hybrid HR happens to be the latest one occupying my wrist. I always have the urge to customize whatever I can get hold of and the McWatchFace was the first in this regard concerning a watch face.

Following that, I moved on to the Fitbit Versa 2 but I could never invest enough time to create a watch face for it.  I had the yearning for a more traditional watch, but ever since I started using fitness trackers, I can't completely wean myself off it. To that end, I got hold of the Fossil Hybrid HR a fortnight ago.

Having never had a Pebble in its heyday, this watch has been a revelation for me in terms of tinkering as it doesn't require much to design something new and share it with the world at large. With that, I present my first watch faces for the Hybrid HR. If you already possess this watch, then feel free to access these watch faces using the Fossil store and let me know if you have any ideas worth implementing.

6 B&W










Review #66: 6 months with Samsung Galaxy Tab S5e ★★★★★

 


Foreword:

The original review was written back in October, about a month after purchase and I have kept it as-is since it still rings true. However, over the past 6 months, with the Covid restrictions, I ended up using my tablet a lot more than my phone and I am still amazed by the value this tablet presents. Probably, that is the reason that Samsung didn't push the tablet hard enough and it is difficult to purchase one in favour of S6 Lite or the S7.

The aspect ratio and colour gamut of the screen lends itself really well to watching videos on it which happens to be a primary use case. However, I will admit that I use the tablet a lot more for reading rather than watching stuff and for that the narrowness of the tablet is an impediment. However, it still does the job and all the reading apps work really well on it, so I cannot complain too much as I knew it was going to be a compromise between watching and reading stuff on the tablet.

Also, I had the opportunity to try Lineage OS on this tablet and did so recently as well with the release of Lineage OS 18.1. While it does the job, the open-source OS still has issues when using proprietary blobs and this is most evident when using the speakers. LOS simply kills the speakers which is a USP for this device. Additionally, the HDMI output on LOS leaves much to be desired in terms of quality and speed. 

In fact, I found Samsung DeX to be surprisingly usable as a desktop interface. My laptop is overkill for simply reading stuff and the tablet with DeX ends up being a great laptop alternative when browsing the web. I have it paired with my Logitech K780 keyboard and MX Anywhere 2S mouse, both of which work really well with the resized desktop apps on DeX. This is not something I thought I would use, but having used it, I prefer having it as an option.

Speaking of DeX, I find that the One UI actually adds value to this tablet through the addition of features like the side bar and slow/limited battery charging options which are genuinely useful. Unfortunately, it comes with a lot of cruft as well, especially due to Knox and as a result I usually go about debloating the OS after every update. But that is for another post as I would like to share how I go about the same.

Lastly, a word about the Spigen Fold cover that I use with this tablet. While it was relatively costly, I found that it has held up really well with all the abuse and the stand aspect of it still works well. It also slots in snugly within the channel on my K780, making typing a chore-free experience on the tablet. In short, Samsung's book cover isn't really a necessity for its exaggerated price tag.

And that's a wrap as far as this device is concerned. It is also good that Samsung has included this device in its new update strategy, so at least security updates can be expected over a 4-year time period since its release. New releases will be less forthcoming but Android 11 is expected to be released for this device in June 2021 and to be frank, I don't think it will matter much as I don't expect One UI to change much, which happens to be the primary differentiator now considering that Android has matured as a platform. Thus, if you can grab hold of one, then by all means go for it, if it fits your budget and use case.

Original Review:

Does it make sense to get a tablet, and an Android one at that? This is a very loaded question and the answer, as always, is that it depends.

I only ever had one tablet before and that was the iPad 3. Back then, the iPad certainly offered a lot more in terms of screen estate, resolution and app experience that you couldn't get on a phone. However, once the novelty wore off, I couldn't really find much use for it apart from reading magazines. It remained stowed away for years until recently when I found that it still works reasonably well (with wonky battery life) and promptly turned it in to a living room clock with weather and news (for which if you are interested, it only consumes 6 Watt).

However, during the lockdown, I found myself using screens a lot more and when unwinding, it usually didn't make sense to take a laptop to bed or to watch or read something on a phone screen. This got me interested in getting a tablet again but I wasn't going to go gung-ho about it and spend a fortune. As a result, I couldn't see myself getting the Galaxy Tab S6 or S7 since it was going to be overkill for my use case.

My use case, if that interests you, was to have something that I could hold easily while in bed and also use for daily reading of newspapers and much more. In short, nothing requiring heavy processing like games. This naturally eliminated the need to have a flagship processor or a refresh rate of over 60 Hz. To be frank, I use a OnePlus 7T with a 90 Hz display and find statements like "can't go back to 60 Hz' to be grossly exaggerated. It may make a difference in games but it is certainly not a deal-breaker as far as scrolling the interface is concerned.

Essentially, among the current mass-market devices, only the S6 Lite and the recently launched iPad 8th Gen fell in to the frame of things. Both of these were priced lower than this device but the S5e offered something that the others didn't. This was of course the great 2K OLED screen along with the 4-speaker setup tuned by AKG. And there you have it, the two features that made this tablet. I would assume some people would have quibbles about the middling Snapdragon 670 and 4 GB RAM, but frankly, they don't matter for reading or watching stuff on the tablet. That is also the reason I am not going to post benchmarks or comment about the camera quality. Every device category is not supposed to be judged by the same metric, but rather by the use case and in this instance, the S5e is not only fit for the purpose, but exceeds it for the price.

In conclusion, if you need a tablet to do something that you cannot do on a phone or a PC, then the S5e is bang for your buck and I wouldn't recommend anything else. True, it wouldn't match up to tablet apps on iOS but I use an iPhone for the iOS benefits and this tablet is there to be tinkered with, and tinker with it you can. It pays to not have all your eggs in one basket, though Apple would suggest otherwise. If you believe that everything has its purpose, then the S5e truly serves as a great tablet for its price and use case.


Tutorial #25: TWEAKING THE ASUS TUF GAMING A15 - Part Deux

 



My previous post about the A15 was about plucking the low hanging fruit of performance. However, there is always the scope of optimising the settings further to gain the greatest benefit for the lowest cost. That is what I was up to on and off since the last post and having reached a satisfactory result, I have decided to share the same for anyone trying to squeeze the little bit extra from this hardware.

The base concept is still the same, to get more out of the GPU at the expense of CPU within the permitted power and thermal budget. To that end, I went through the process as follows:

1. Reduced the CPU temperature and Normal/Short/Long TDP limits to 85 and 25/35/45 respectively in Ryzen controller to provide further headroom to the GPU.

2. Reset the MSI Afterburner settings to stock which had the following curve for my RTX2060.

3. Ran an actual game as I would like to play it (in this case Dishonored 2 at 2K Ultra with HBAO and Triple Buffer) and noted the Average and Maximum GPU frequencies attained in the middle of the game.

4. Few observations first. Neither the CPU or GPU are thermally throttled in any way. Instead, the GPU hits the power limit which in the case of the A15 is 90W. Note that the included GPU is the RTX 2060 Notebook Refresh and thus it is a 110W TGP part. This indicates that the laptop does have thermal capacity to spare, especially as I had conducted these test with an ambient temperature close to mid-30 degree Celsius. Having the option to push the GPU Power further would have been great but with that being an impossibility with a locked BIOS, the next step was to figure out how to extract the most from the hand I have been dealt with.

To that end, I noted the frequencies which yielded the sustained performance (1560 MHz @812 mV) and the peak performance (1755 Mhz @918 mV) in-game.


This concluded the stock performance analysis. Now, there might be multiple guides present that put forth different suggestions as to how you can proceed with undervolting or overclocking but I decided to use these figures to try to set a target that I wanted to attain. In this case, it was to try to push the stock sustained performance to the lowest voltage (i.e. 1560@700) and the peak stock performance to the sustained voltage (i.e. 1755@812). Doing so manually with a smooth curve was going to be quite a challenge, so I decided to take a bit of a shortcut in attaining this objective as follows:

5. Executed the Nvidia OC Scanner within MSI Afterburner to produce an OC curve. The curves are not always the same, so I executed it a few times, also at slightly different CPU TDPs to come up with the curve that resulted in the highest boost frequency. In this case, it was as indicated below.


If I look at the frequencies at the concerned voltages, then it is 1515@700 and 1755@812. Thus, it seems I have almost attained the target I set out for without doing much.

 
Taking a look at the HWInfo figures again with the OC curve, it can be seen now that the sustained frequency has jumped to 1725 MHz from 1560 Mhz which is a decent OC. Also, the peak frequency now is at 2040 MHz which is an even bigger leap but it comes at a much higher voltage (1006 mV). The effect of this however is that the GPU is now hitting all the performance limits apart from the thermal one.

Almost there, but "almost" is not good enough, so I had to push it a bit further.


6. At this point, I decided to try to move the curve to the left, in effect overclocking the OC curve even further to see how much more performance can be extracted from it. I started by essentially shifting the curve to the left by 25 mV but as soon after I started encountering artifacts within the game indicating that I had pushed it a bit too far. As a result, I shifted the curve by 12.5 mV instead and found it to be perfectly stable. 

The other change I did was to flatten the curve at the half-way mark of the complete voltage range which is at 975 mV. There are various reasons to do so, primary of which was that the GPU never really reaches the frequency associated with that voltage and if it does as stated in the point above, it is for a fraction of a second. Consequently, it also saves the effort of manually adjusting the curve in futility. An argument could be made that the curve can be flattened even earlier to essentially attain an undervolt but I wanted to allow the GPU to boost to its practical maximum as much as possible.

With the above, after smoothing out the double frequency jumps (15 Mhz instead of 30 Mhz for a single increment in voltage step), I was left with the curve indicated below.


It starts at 1560@700 (surprise!), reaches 1755@800 and peaks at 2010@975. So how does this curve now fare within the game?


Firstly, we are back to only hitting the power and utilisation limit. The sustained in-game frequency is now 1755 Mhz, a further 30 Mhz boost from the default OC Scanner curve. The peak frequency is now 1965 Mhz though as against 2040 MHz earlier, but as I mentioned previously, it is transient and if you look at the average GPU power, it has come down to 69.4W compared to 70.3W for the OC scanner curve and 71.3W for the default curve. Amusingly, the maximum power consumption was over 97W with the stock curve and I also observed it breaching the 100W barrier in an intermediate test. May be it is due to some quirk in HWInfo or otherwise, the card is indeed capable of going over its locked TDP of 90W in some cases, though without much benefit.

7. With the GPU OC sorted, next, I wanted to see if I can push the CPU a bit more in co-ordination with this curve. You will have to take my word for it, but I tried increments and decrements for all the TDP values while keeping the temperature limit at 85 degrees Celsius and I finally found the best performance at Normal/Long/Short TDP of 25/40/50 respectively.


The proof, of course, is in the pudding. Thus, I present to you now, the comparison between the stock performance and after the CPU/GPU tweak. The duration of the HWInfo figures spanned from the launch of the Time Spy test to the calculation of the score.

Stock:

Post tweak:

A good jump and also a slightly higher score than the tweak in my last article (6703). What it doesn't indicate though is that the power consumption is lower than last time.

8. One last thing! I didn't at any point mention anything about the GPU memory overclock because I kept it for the last. After trying out different increments, I settled for a boost of 500 MHz as it was stable and didn't lead to any noticeable increment in power consumption and thermals. With that, here is the final result.


To put things in perspective then, this is how the tweak stacks up against the stock setting.








Tutorial #24: Tweaking the Asus Tuf Gaming A15


Previously, in my review of the laptop, the only tweaking I had undertaken was an auto-overclock of the GPU which, as per expectation, yielded a performance improvement of around 6% overall with only a slight loss in CPU performance, purely on the basis of the additional available thermal headroom.


During that time I had left the CPU untouched because AMD does not officially support tweaking on laptops and Ryzen Controller did not work for me then. However, later I came across Renoir Mobile Tuning and found it to be operational for this laptop, albeit with a few bugs. I switched to Ryzen Controller again and found that it too now worked well for Renoir with the additional benefit of applying the setting automatically on boot.

With a CPU tuning tool in place, the next thing was determining what to do with it. While these tools often end up as overclocking utilities, my intention couldn't be further opposite to that. The idea was to effectively underclock the system without losing performance i.e. to reduce the temperatures while still maintaining a performance boost over the stock settings.

To cut a long story short, I played around various combination of settings to finally settle on one that seems to work the best. Not that it an exhaustive analysis but rather the most practical among the ones I had tried. Note that I only experimented with the Boost TDPs and the temperature limits. The boost duration seemed pretty logical and I did not want to introduce yet another variable that muddied up the testing. Eventually this resulted in the following changes:
  • Temperature Limit: 90
  • Long Boost TDP: 54 
  • Short Boost TDP: 50
For reference, the default temperature limit is 95 with long and short boost TDPs of 60 and 54 respectively. Also, I auto-overclocked the GPU again to make the most of any benefit available from reduction of the CPU performance. So, how did this theoretical reduction in CPU performance impact the benchmark scores for Fire Strike and Time Spy compared to the ones from the review?


As expected, this has quite an impact on the CPU performance as it has dropped by nearly 5% but on the other hand the graphics score has jumped by 1% resulting in an overall gain of 0.7% on Fire Strike, taking it past 16,000 for the first time. However, the result for Time Spy was more interesting as there was a minor loss instead overall indicating the underclock has more of an impact of DX12 than it does on DX11, which is probably not unexpected. Note that this is an indication of the gain over the gain already achieved by overclocking the GPU originally, so overall the incremental gain is still worth it.

Lastly, the laptop has a secret weapon up its sleeve. Until now, all the tests were conducted using the default Performance mode. However, there is also a Turbo mode which sets the fans whirring to possibly the maximum setting under full load. Yes, it boosts up the scores even further. Below I have again attached a comparison of the Turbo mode performance for the stock CPU settings in comparison to the underclocked one and it is quite the same as earlier. While the DX11 performance is higher with the underclock, it is lower by an equal proportion in case of DX12. 

It has to be kept in mind though that apart from the scores, the underclock has an additional benefit in reducing the overall temperatures and also prolonging the life of the components. Also with the combination of the 4800H with the RTX2060, it is the latter that is going to hit the limit rather easily compared to the former, so a sacrifice of CPU performance for a GPU gain makes a lot more sense.

Finally, I leave you with a comparison of the current profile comprising of a GPU Overclock and CPU Underclock on Turbo with the stock GPU and CPU settings.

A jump of 7.8% on DX11 and 6.6% on DX12 with lower overall temperatures to boot is nothing shoddy. Seems something called as free lunch does exist after all.