Posts tagged with “benchmarks”

Tale of Two Laptops

Last night, I traded my iPad Pro for a Huawei MateBook X Pro (2018) (hereafter "matebook") and now I'm going to spend some time comparing this laptop to my current laptop, a Panasonic Let's Note CF-SV8 (hereafter "SV8"). Both machines are very similar in some ways and incredibly different in others, so I thought I might take the opportunity to write a little bit about this. This is not a review of either machine and is probably a bit premature as I've had the SV8 for a few months and the matebook only a few hours.

Lets start with the objective boring stuff, the specs.

SV8 (CF-SV8RDAVS)MateBook
CPUi5-8365U 4c/8t 1.6GHz Base, 4.1GHz boost, 6MB L3, 15wi7-8550U 4c/8t 1.8GHz Base, 4.0GHz boost, 8MB L3, 15w
RAM8GB LPDDR3-2133 (soldered)16GB LPDDR3-2133 (soldered)
GPUIntel UHDIntel UHD 620 & Nvidia GeForce MX150
SSD256GB m.2 NVMe 2280512GB m.2 NVMe 2280
Other StorageDVD-RW & SD XC slot-
Ports3x USB3.0, USB-C Thunderbolt 3, HDMI, VGA, 1GbE Ethernet, Headset, 16v power1x USB3.0, USB-C Thunderbolt 3, USB-C 3.0 (data & power), Headset
NetworkingIntel I219-LM Ethernet, Intel Canon Point-LP CNVi Wireless-ACIntel Wireless-AC 8275
Screen1920x1200, 12.1"3000x2000, 13.9", capacitive multitouch
Dimensions283x204x25mm (11.1x8x1in)304x217x15mm (12x8.5x0.6in)
Weight1Kg (2.2lbs)1.3Kg (2.9lbs)
Battery43Wh (5900mAh 7.2v) removeable56Wh (7410mAh 7.6v) fixed
OtherIndicator LEDs, Windows Hello camera support, full magnesium chassisBacklit keyboard, fingerprint reader
Retail Price (Approx.)$2,200$1,700

So, what does this tell us? Well, these are both "thin and light" medium/high class machines from 2018, though their target audiences are very different. The SV8 isn't even sold in the US, and is targeted squarely at Japanese businessmen wanting a high end and very practical Windows laptop with good connectivity and great battery life. The matebook, on the otherhand, is sold worldwide and is targeted at people who want a Mac but want to run Windows on it for slightly less than Apple prices. It's a high end machine, but not really suitable for "traditional business use" and it suffers from the port-deletes found on most similar thin and light laptops today.

Aside from the differences in CPU and RAM (which, I think it's worth mentioning - the SV8 is available in an i7/16 config, and the matebook is available in an i5/8 config) the largest difference between the two from a hardware perspective is that the SV8 has an integral full-size optical drive, which, for a 12" laptop in 2018, is nothing short of incredible. It's not something I need or use, and in fact I have it disabled in the BIOS, but I can understand how some see the appeal. Optical drive delete models exist and are approx. 100g lighter, but they don't seem to be very common. Additionally, the matebook includes a dedicated Nvidia MX150 GPU. This is a GPU suitable for "light gaming" (eSports Titles and the like) and something I would just as soon not have. It's not possible to disable entirely in the BIOS, but it's trivial to do so in both Windows and Linux. Doing so, of course, increases battery life. Personally, I find the inclusion of a dGPU for occasional 3D tasks to be more beneficial than an optical drive. I can plug in a USB optical drive, but eGPUs are decidedly more of a pain.

I don't use a laptop as my primary machine, so lots of RAM and a powerful CPU are much less relevant to me. The SV8 has performed every task I have cared to throw at it handily, and while I haven't done as much on the matebook, I expect it to perform just as well. I have benchmarked both machines, both with Geekbench 5.3.1 and an xz compression test if the reader is interested in raw numbers.

Both machines have beautiful screens. The matebook's panel is glossy, however, because it's also a touch screen, which is not something I care for. I don't mind glossy panels, but I know that some hate them. I rarely, if ever, use a computer outdoors, and I try and avoid situations with bright overhead lighting when I can. 3000x2000 at 13.9" on the matebook is far too high a DPI for me to use at 100%, but 200% is too large. 150% feels just right in windows, but fractional scaling suffers considerably under linux, and may even be impossible when using the Nvidia GPU. I would prefer to run my display at 100% under all circumstances, but on the matebook it's just not possible for me without reading glasses. The SV8's panel, by comparison, is 1920x1200 at 12.1", and I find it very usable at 100%. There are times when my eyes are a bit more "tired" and focusing on the smaller characters is a bit of a struggle, but this can quickly be resolved by blowing up my web browser or terminal to 125-150%. Both panels are better than a typical 1920x1080 display due to the added vertical height. The matebook is 3:2 and the SV8 is 3.2:2. Both are wonderful, but the added height on the matebook wins for me.

On the question of input devices, it's no question that the matebook wins for me. The keyboard is wider with larger keys, and it has a more standard American layout that I am used to elsewhere. The keys have good travel and feel, and nothing wobbles or bends when typing. It's not the SV8's fault for having a JIS keyboard, as it is a machine primarily sold in Japan, rather it's my fault for using a JIS keyboard in an American layout. The SV8's keyboard has many keys that are specific to inputting Japanese text, which are unused by me, and I would prefer they not be there at all to make the other keys larger. There is a configuration of the SV8 that is sold in Singapore which has an American style keyboard, but I have not been able to find one of these yet. I also suspect if I had smaller hands I would like the SV8's keyboard more. The SV8 keyboard does win out in a few ways, though: it has a contextual menu key, which I use daily, it has dedicated Insert/Delete keys (the matebook only has delete), and the arrow keys are in the "normal" inverted-T configuration, rather than the mushed macbook-style that the matebook adopted. The feel of the SV8 keyboard is "fine", on the mushier side but not bad. It also feels very rigidly affixed to the case. The matebook's is also backlit, which I appreciate; the SV8's is not.

I almost always use a bluetooth mouse when using a laptop on a table, but I'll use the trackpad plenty around the house or when traveling. The matebook has a gigantic glass covered trackpad, much in Apple's style, and it feels great and is very responsive. The SV8's trackpad is also very responsive, and it's unique circular shape is interesting, but it is very small, which can be frustrating at times. The dedicated buttons of the SV8 are nice, but missing them on the matebook isn't a dealbreaker. Something I wish both had was the ability to turn off the trackpad from a function key. If I'm using a mouse, I often don't want any input from the trackpad.

The matebook's thinness causes it to loose the I/O battle without question. A single USB3-A, two USB-C (one of which is TB3) and a headset jack. That's it. Not even an SD card slot. The SV8 on the other hand comes in with a full compliment - three USB3-As, a USB-C for TB3, HDMI and VGA, ethernet, and a dedicated barrel jack for power. The SV8 also has a full compliment of indicator LEDs: power status, lock keys, and activity of HDD and SD while the matebook only has a charging LED on the side. USB-C charging is an option for both machines - the matebook requires it with no alternative, and the SV8 supports it as long as the machine is either running or in standby. For some reason it won't charge from USB-C if it's powered off - I haven't figured out why. Both support Windows Hello if you're into that sort of thing - the SV8 has the IR camera while the matebook has a fingerprint reader on the power button. I don't use either. The webcams on both are good, though the SV8's is clearly better, full 1080p. The matebook's is 720, and it pops up out of the keyboard between the F6 and F7 keys, which is a very peculiar feature. It does enable the screen area on the matebook to completely fill the lid, which is very attractive. The matebook has fairly full-sounding speakers which can be incredibly loud, which is something I find frustrating. In Windows I find the 10-15 volume level to be appropriate. The SV8's speakers are "fine" but definitely not something I would want to use for listening to music, contrary to the matebook. I haven't used the microphones on either, but I imagine they're both pretty bad, but acceptable in a pinch.

I don't know how Huawei's batteries track over time, but I don't have a tremendous amount of confidence in it's long term capacity. The SV8, on the other hand, I expect to retain near-original capacity for years, as has been reported by other Let's Note owners. Panasonic is just really good at batteries, go figure. I have never needed to go more than about 5 hours on a charge, and both machines will do so easily. The 43 watt hour battery in the SV8 is reported to last up to 13 hours under light duty loads, and I can believe that. The matebook is advertised as lasting 12 hours, which I do not believe. Reviewers have placed it at the seven to nine hour mark, which as I have stated is plenty enough for me. I don't know if I will be able to get a replacement for either, should I ever need one.

So, aesthetics. The matebook is boring. Lots of people will probably find it to be very sexy and attractive, but it's just a 13" MacBook Pro clone. It does a good job being a MacBook clone in that it feels very solid and the fit and finish is excellent. Pretty, but boring. The SV8 on the other hand is a cute little pragmatic powerhouse that many people find to be tremendously ugly. It's true that the aesthetics of Let's Note laptops haven't changed significantly in 20 years, but Panasonic is clearly taking a "if it ain't broke don't fix it" approach to their laptop design, and I am thoroughly okay with that. The chassis is composed entirely of magnesium, so it's very strong and light. The lid has an interesting wavy design in the top that is designed to distribute weight that might otherwise crack the display. When you pick up the SV8 you are immediately surprised at how light it is, and you suspect it's made of thin plastic like a netbook, but it's not! The SV8 also has a top mounted hinge, which I greatly prefer to the matebook's macbook-style wrap-around hinge. The matebook is "sleeker" though - thinner with no protrusions. The SV8 is much thicker, and the way the battery and feet stick out from the bottom causes them to often get caught on something when taking it in or out of a bag. Additionally, the strange footprint and thickness of the SV8 makes finding a suitable small sleeve or case a challenge, whereas the matebook is much more of a generic 13" laptop.

Conclusion? I don't have one yet. They're both good laptops, and for the foreseeable future I'll continue to use both in varying roles. I look forward to traveling with both when Global Pandemic is over as I would like to have more time with each on the road. I can see the SV8 being more of a desk queen, living beside my desktop in a very dignified role. The matebook, on the other hand, is a beater - if it falls off a table or my daughter spills chocolate milk on it, well, that's why I have an accidental damage rider for laptops and tablets on my homeowners insurance. I do not know if it would be possible for me to get another SV8.

SV8 and Matebook

SpeedStep® Limiting Single-Threaded Tasks

This is something that will need to be further fleshed out by benchmarks and research, but I wanted to start writing about it now, before I forget. More is certainly to come.

Context

First, a little backstory for context. The company I work for, Esri, just released version 10.6.1 of their ArcGIS Enterprise product. Something I found very appealing about this release is that it includes a new interface and set of tools for processing imagery from UAVs into orthophotos, DSMs, and DEMs, called Ortho Maker. Esri has a desktop product that does this as well, Drone2Map, which is based on the Pix4D Mapper engine. Ortho Maker is unique in comparison because it is intended to run on server infrastructure rather than workstations or HEDTs, and anyone who knows me will know that I find that incredibly appealing.

So, when 10.6.1 was formally released, I wiped the TS140 'goonie' to a clean slate and configured it as follows: 2x8GB RAM (I moved the other 2 to TS140#1), 256GB Samsung 840 Pro boot/primary data drive, 2TB WD Green as a backup/scratch drive - this simply because it was laying around and I figured I could use some extra storage that didn't need performance or redundancy. Finally, a fresh copy of Server 2016 - which took forever to update, then a base enterprise deployment of ArcGIS Enterprise 10.6.1. Now, this also required an Image Server license because raster processing tools are used in order to generate the final products. In a production environment the server with the image server license would not be the same as the server running the portal and hosting server (you can read more about ArcGIS Enterprise architecture here), but since I am the only individual that will be accessing this server, and I know that nothing will be putting load on the hosting server or portal while the image server components are working.

I have a few different sample UAV imagery collections that I use to test this software, both my own and others that are available publicly on the internet, so I began with processing a few to get a better feel for the capabilities of the software. As a solution engineer, part of my job is to thoroughly test all the components of our software that I could find myself recommending to a customer, and this is no exception.

Where SpeedStep comes in...

After watching Process Explorer and the Windows Task Manager while the various stages of the ortho maker's processes were run, it became clear that these are single threaded tools. I imagine this will change in future releases, and there may be an option to make it multi threaded in the future, but as it stands each of the tasks only runs one thread at a time - each process never taking more than 25% of the total CPU time (the E3-1225v3 is 4c/4t).

Windows Task Manager shows the current clock speed of the CPU, which is very useful in cases like these, and what I was seeing was clock speeds much lower than expected. The E3-1225v3's base is 3.2GHz, with boost up to 3.6GHz, however Task Manager was showing that the system was rarely, if ever, hitting higher than 1.2GHz. Now, we all know what SpeedStep is - it's that fabulous technology that changes CPU speed on the fly, keeping power consumption and temperatures down. The dark underbelly to SpeedStep, apparently, is that it uses total CPU usage to gauge whether it needs to crank up the speed rather than per-core usage.

In this case, that's a problem. If the server is running two Ortho Maker tasks, and is otherwise idle, the CPU will only ever be at 50% + a small amount of system overhead, maybe 15%. What that means is that the system will never determine that it's necessary to step up, and the performance of those single threads will be far less than they could be. I suppose if I was concerned about temperatures and fan speeds then maybe that would apply, but this is a server we're talking about here.

How to resolve it

Thankfully this is an easy problem to solve. Disable SpeedStep. This can even be done from within windows, without rebooting. Launch powercfg.cpl and change the power plan from "Balanced" to "High Performance". If you want to see what this does, click "Change plan settings" then "Change advanced power settings" - under processor power management both the minimum processor state and maximum processor state will be set to 100% - meaning that in my case the CPU will always be at 3.2GHz, and will boost when possible.

SpeedStep can also (usually) be disabled from the BIOS, and this would be my recommendation. I'm currently working on this server remotely so that's not an option for me, but I will be addressing it soon!

What other people are saying

Not surprisingly, quite a few people have discovered this quirk, and Microsoft even has some KBs that address this when it comes to using Hyper-V. Basically what it comes down to is "know your workloads, and know your hardware". In many cases, leaving speedstep enabled will be best.

What I need to do next

More research, and some single threaded benchmarks of my own. I also want to test this theory across multiple machines - especially the C220M3 (currently in storage...).

Resources

New CPUs! (And more?)

The CPUs and heatsink actually arrived on Friday, two days ago, and I'm just now getting around to writing the post. I don't have a whole lot to say about the upgrade process, it was pretty typical. I only really ran into two snafus: 1) I slightly dropped one of them while placing it in the socket, and the corner fell right on the pins. Normally I wouldn't really be worried but these E5s are rather big and heavy, and two-three pins were definitely bent. Thankfully, with a needle and magnifying glass I was able to easily realign them and the CPU has registered just fine. Also thankfully it was the second CPU socket, so in the event that I trashed the pins, at least the first socket could still work (AFAIK it's not possible to put a CPU just in the second socket of these boards, and I'm not too keen to find out anyways). The 2)nd snafu was that I didn't realize that hyperthreading was disabled - presumably because the precious CPU was just 4c/4t - so all of my initial benchmarks are useless.

Anyway, pictures!

First was the way the heatsink seller packed the heatsink. It's in this very nice little enclosure of cardboard and styrofoam. I have actually received heatsinks in the mail that were crushed slightly and had bent fins, so seeing this is nice.

And here's all the bits laid out on top of the chassis pre-upgrade: test!

Aww yiss...

Now for the "And more?" ....

Well, apparently, my ESXi license only allows me to allocate 8 vCPUs per VM, which just wasn't going to cut it with 24 available. I should have known better than to configure this machine right away as a production environment, because OF COURSE I would want to play with it in different configurations with different OSes. So, what I did was re-configure my Enterprise GIS server (SFF ThinkCentre M91p) with Ubuntu 16.04 so it now runs caddy, observium, UNMS, (other misc. docker bits), plex, my SSH bouncer, and others I am probably forgetting. This frees up the C220 M3 to be more of a playground.

The first thing I did was pull the two 128GB SSDs and the 750GB laptop drive, leaving just the 600GB SAS and the two 480GB SSDs. I also pulled the Adaptec 2405 and switched back to using the onboard RAID controller, which can apparently be configured as either intel or LSI softRAID. Server 2012r2 didn't see the intel RAID arrays, but the LSI ones worked fine, so that's what I'm using now. I have setup the Hyper-V role but have not tested it out much yet, so that's next. I have done some remedial benchmarking and overall it comes out a bit ahead of the ML350 G6 - not significantly, but this is to be expected. The X5660s in the G6 have a 400MHz higher ceiling, but the ML350 G6 also pulls about twice as much power (both under load and at idle) than the C220M3 does.

More testing and fiddling with Hyper-V is forthcoming.

First Wave of Disk Benchmarks

First off, I will heartily admit that these benchmarks are by no means good, I just wanted to do them quickly. They were performed in Arch with the gnome-disks tool's built in benchmark.

All this tells us is "striped SSDs are faster than single HDDs!" What I really should do is get some good (and identical) HDDs and perform the same test, and test the SSDs standalone as well. I don't want to fiddle with it that much, so this is what we have.

It would be better to run these tests bare metal, not through a VM/Host, but seeing as I am going to be using this machine solely for VMs, it made sense to me to know what the VM performance is going to be. I don't really care about the absolute performance, since all it will tell me is how much overhead ESXi is injecting into the situation, and I don't really care about that.

In any case, I think my RAID1+0 of four 240GB drives will perform great.