*** Official Clevo P65xSA/SE/SG / Sager NP8650/51/52 Owner´s Lounge ***

Having read and reread the manual myself, I agree that the B-keyed slot almost certainly supports SATA only, and not PCIe x2...

We also need someone to test and be sure that a 3042 M.2 LTE card will work in the B slot with a PCIe x4 SSD in the M slot, and to see if the SIM slot is functional or indeed just a block of foam.

Not quite what I said... but yes, if such a device existed, it would have to be PCIe or USB, not SATA. And I did say it would have to have the A key...

 

Yeah, kinda makes sense. Cause as per current design, it is either 2x7mm (14mm) or one 9.5mm. With addition of 4mm, we can have one 7mm and one 9.5mm (theoretically) and 2x9.5s would not be possible as it would require 5mm addition.

 

I would still be concerned with two drives stacked on top of each other. Probably cool running HDD on bottom, and SSD on top for most airflow.

 

Do HDDs really generate LESS heat than a SSD? I would have figured it the other way around.

 

HHDs generate less heat I'm 99% certain unless there is something I havn't heard about. I believe Wing is just talking about how to configure them to get the best airflow for the HHD which is hotter(?).

 

Indeed, this is something I've been ... "correcting" ... for years now.** A typical 1-3 platter 2.5" drive only consumes 0.4-0.8A at 5V, for a typical 2-4W. Back when commodity SLC NAND devices were much smaller, in the 30-60GB range, they typically consumed similar.

**SIDE NOTE: An added "oversight" that I regularly had to "correct" was that people were comparing 2.5" NAND devices to 3.5" drives. In the case of 3.5" drives, they always consume 1-2A@12VDC (easily 12-24W, maybe a 0.5A/6W "green" mode only part of the time), instead of similarly sized 2.5" platter drives (which are 2-4W max during operation).

But nowdays, NAND devices are blowing well past 1A, and even 2-3A at larger sizes (5-15W, often 6-8W draws at times). This is why a lot of people have trouble with putting 2.5" NAND devices in external, USB powered enclosures, even when using a 2nd USB port for power. It's not only for the NAND write requirement (which is vastly more than reads -- usually at a completely different voltage), but the increased amount of DRAM buffer. DRAM cells are grossly inefficient, leaky semiconductor designs.

With anything new, it's best to work with your OEM who is going to have the first samples and sales of the components they support, especially one like MythLogic that assembles in-house.

When the P65x has been around a good 6 months, then there will be a lot more user experience on forums like here. Until then, I'd use my OEM as a crutch, especially for that crucial first year of warranty, let alone the first few months. If a PCB or other component is going to fail, or have various integration issues, it's usually in that first 30 days, especially for a very new design.

The overwhelming advantage of NAND over platter is random-access. Platter's seek times during random access are what bring down an even nearly 100MBps sequential transfer rate to only 1-5MiBps. NAND will regularly break 25MiBps, if not 50MBps, under the most inefficient of allocations because it is far more of a random access solution. That's why it's at least an order of magnitude faster than any platter in typical usage -- from boot to launching applications and reading large files.

I'm going to move my existing 2.5" platter over to my P65x when I get it. However, I'm purchasing the 32GiB of memory, 512GB M550 M.2-SATA and everything else with system. I prefer to pay the 10-15% "premium" to get it from the OEM, all tested and warrantied. But that's just me. I mean, in the ending, saving 5% on the total system cost is not really worth it.

SIDE NOTE: I have the same attitude on OEM v. aftermarket batteries, at least when the unit is new (within 2 years). I prefer to go with the OEM's certified Li-Ion, rather than save 10-15% of the overall system cost. In the case of dSLR's, I am adamant about this, even 5 years down the road ... because it's something I hold up to my eyes. There is no "generic" and "universal" Li-Ion out there that everyone designs to -- otherwise we'd all be using Rechargeable CR-V3 batteries (which vary wildly in spec).

If your CPU only supports dual-channel, like the Intel i7-4170HQ and similar in the P65x series, then I highly recommend only two (2) DIMMs to start, one (1) per channel. When you add more than one (1) DIMM per channel, you introduce additional "noise" (I won't go into EMF/EMI). By default, most JEDEC specifications reduce the clock (so any SPD on the DIMM will provide such timing during POST) when more than one (1) DIMM is used per channel. It's not that much of a difference, because the latency of the DRAM doesn't change (timing will be lowered/better for reduced clock, so it evens out, latency-wise), but it still never hurts to keep noise minimized.

By going with (2) 8GiB DIMMs, you'll minimize noise and be able to upgrade to 32GiB in the future as well -- win-win.

 

I have a question for those of you with the Clevo P65x series already in your hands ...

Given the Intel i7-4710HQ and both the Intel QM86 and Intel QM87 chipsets all support VT-d (not to be confused with VT-x), I just want to verify the CPU feature flag shows up in the OS as well -- whether there is an option to enable it in the firmware or not (e.g., it's just always enabled).

Spoiler

Intel is notorious for disabling VT-d on most consumer desktop chipsets (e.g., anything but the "business" Q87 chipset in the 80 series), although several OEMs have been enabling it, despite Intel's requirements that it be unsupported on other chipsets (e.g., even the high-end Z87 in the 80 series).

A bit of my focus right now in OpenStack is Network Function Virtualization (NFV) and things like VT-d in the notebook would allow me to not have to lug one of my SFF Opteron 4256 systems around just to test CPU-mem-I/O pinning with cgroups on the KVM host/guests, etc... It's used for more than just VMware ESX (although that's the common request/desire for VT-d support in a "consumer" board).

 

B.J., that is one beautiful Vette. I always remember this video when I see one. A good chuckle for you, hopefully.

 

OT ...

Spoiler

Thanx. I was partial to the Jetstream Blue Metallic, but the dealer was trying to dump that '09 Z51 Coupe with the Cyber Gray Metallic (the new color introduced in '09 for the production ZR1), as the new '10 Z52-Grand Sport models had come out. My wife preferred the color, and it uses virtually the same metallic as the Blue.

I.e., it's a dark gray that, in the sunlight, has a deep, bright blue tint and shine to it. It never fails to catch everyone's eye when the sun is out here in Florida. One of the guys in our club liked it so much he got a '10 ZR1 with it.

Because GM's source for the Cyber Gray Metallic is the same as for some Toyota and other sedan models, I've found I get the most "votes" at car shows from women, instead of men. In fact there was about 18 months where you couldn't order it back during the Japanese quake situation. There's only one factory that makes it.

Despite always being a top 3 color from '09-14, GM dropped the color from '15+. They've gone with a lighter gray with a far different metallic now, likely for single source avoidance.

I don't know exactly what it is, but there are way too many "kids" out there with used, stock C5s they picked up for $20K. They think they can just floor it and do all sorts of stupid crap. It's one of the reasons I don't do a lot of "hot laps" on course, let alone "accelerate fast" on the road, and "just let the C5 punk win." That's probably what the C6 driver (left) should have done here, as the "C5 punk" (right) lost control.

My wife and I attended the Corvette World Tribute at Road America back in 2012, and there were lots of C5s just cutting me off and sometimes doing it wildly. Over a good 90 minutes of hot laps over two sessions, I must have almost been hit three (3) times, all by stock C5s with a young driver.

Otherwise any recent model Corvette (like '05-'13 C6s) continues to be the cheapest car in the US to insure, as long as you're aged 30+ and don't have too old of a year (like a C4 or C5) where younger kids are buying them and knocking up the crash statistics. It has a value of 3x of any of my other cars, but the overall liability+comprehensive still makes it the cheapest of all my cars to insure. Other than a $200 sensor going in early 2013, which caused the active handling to kick in aggressively, I haven't had an issue with it over the 5 years and 31,000 miles I've put on it.

If I ever do race my '09 Z51 ( if I don't sell it for a good price), likely after I get a C7, it will be at a High Performance Driving Events (HPDE), with non-amateurs. I did Bondurant at Firebird Lake in 2012 and will likely do Fellows at Spring Mountain next year, and eventually get my full SCCA classification. I just don't have time, and I really need to drop some weight too. I'm not in good shape at age 40+ for 60 minutes of constant 1G+ lateral. To date I haven't autocrossed it and only done a few "hot laps," so it's been babied. It's just been sitting in the garage for most of the last 2 years (I've just been too busy). Haven't even a thousand miles over those 2 years, and probably will for the next 5 too.

 

Yea, I ended up going AMD for my latest desktop build just because getting VT-d was going to cost me an extra $300. Really bugs me that they do that.

As for whether or not it's enabled -- it probably is. The it worked fine in the W230SS (4710MQ, not HQ). If it isn't, it's probably something Prema will bake into his hacked BIOS.

I didn't know pinning required VT-d, at least for CPU. Can you explain the concept of 'mem' pinning? I definitely have some annoying memory speed issues in kvm and would be interested in anything that could fix them. For the moment I'm forced to use hugetlbfs, which, frankly, sucks. Transparent hugepages is enabled and working but gives awful performance (maybe TLB thrashing? as I'm forced to allocate memory for hugetlbfs at boot and more likely to get large sequential blocks of memory, which is probably easier on the TLB?).

 

Memory options


I've got quite confused with the talk about SSD options and would appreciate some advice.

PC Specialist offer various options.

I was originally going for option 1 below.. but now I'm not sure this would be the best after the recent discussions.

I'd was planning on using a traditional HDD for storage and an mSATA for Windows installation etc.



Option 1

HardDisk
1TB WD SLIM BLUE WD10SPCX, SATA 6 Gb/s, 16MB CACHE (5400 rpm)
2nd Hard Disk
NONE
mSATA/M.2 SSD Drive
Plextor PXG128M6e 128GB M.2 SSD (upto 770MB/sR | 625MB/sW)


Option 2

Hard Disk
1TB WD SLIM BLUE WD10SPCX, SATA 6 Gb/s, 16MB CACHE (5400 rpm)
2nd Hard Disk
250GB Samsung 840 EVO SSD, SATA 6Gb/s (upto 540MB/sR | 520MB/sW)

Option 3
HardDisk
1TB WD SLIM BLUE WD10SPCX, SATA 6 Gb/s, 16MB CACHE (5400 rpm)
2nd Hard Disk
240GB KINGSTON HYPERX 3K SSD, SATA 6 Gb/s (upto 555MB/sR | 510MB/sW)

Option 4
Hard Disk
1TB WD SLIM BLUE WD10SPCX, SATA 6 Gb/s, 16MB CACHE (5400 rpm)
2nd Hard Disk
240GB KINGSTON V300 SSD, SATA 6 Gb (450MB/R, 450MB/W)

Cost wise there is little difference.

Thanks

 

Go with option 2.

 

Yeah, when I can buy 3+ year-old AMD SR5600 mainboards with low-power (35-40W TDP) Opteron 4000 series 6-8 core CPUs for a few hundred, it's difficult to buy any Intel product. Heck, even the Jaguar/Kabini and newer Puma/Beema ULV solutions are far more feature rich than the Intel Celeron/Pentium ULV options, and a lot cheaper too.

More and more OEMs are starting to enable it, despite Intel's licensing requirements on various chipset releases. I won't go into my tirade on Intel purposely crippling it's low-power i-series products (especially the Celeron/Pentium branded i-Cores) to avoid canniablizing their God-awful Atoms which just don't perform (let alone they are now rebranding Atoms as Celeron/Pentium too now).

The rest I'll block as "Spoiler" as it's more OT from the Clevo ...

Spoiler

I probably shouldn't have over-simplified, because pinning is just one small part of the equation. It's largely for I/O.

I/O latency is everything in a trading, telecom, etc... environment. One needs to be able to ensure a guest can most directly access memory mapped I/O, and fill it as soon as interrupts are raised, etc... That needs to happen, even for single socket.

In the case of multi-socket with today's common NUMA designs (whether Intel QPI/DMI or AMD HT), the I/O needs to always be done on the CPU and to/from the memory where the I/O bridge is most local. And even single socket is increasingly almost NUMA-like with how cache and memory is partitioned between cores too.

So ... in the virtualized world, things like VT-d and I/O MMU support, along with SR-IVO, needs to be mapped in both the host and to the guest, proper. In Linux we've long used cgroups, and this now means we're trying to assign (auto-assign, if possible) the VM with the I/O sensitive applications to the cgroup that makes the most of the topology.

In fact, this is a big focus for systemd (which always sets up cgroups -- at least if the feature isn't disabled -- by default) long-term.

What issues and what applications?

And before I start ... I have to ask ...

You're not setting vm.swappiness = 100 in your VM guests (regardless of OS) like EMC/VMware's own tuning guides recommend, correct?

I literally want to endlessly smack the smelly chocolate out of the person at EMC/VMware that started that non-sense. That setting is designed to give major over-commit so you can claim 20:1 (or whatever) VM:Host "cost-benefit savings" to finance, and will absolutely result in self-destructive performance, especially on KVM. Remember, KVM is very Linux kernel-integrated, so there are inter-host/guest aspects that are wholly different than for ESX.

Also let me know if your VM performance issues are because they are running Oracle SQL, which is a whole other story.

The main issues I've really had with KVM have been related to how poorly .qcow2 image files work compared to DeviceMapper organization (e.g., LVs). I never use .qcow2 for production images, only LVs. This is also what RHEV always uses too, unless one exports an image.

The only other thing I can think of would be Virtual Memory-related, especially buffering and not waking up or forcing pdflush to act. I've long had this issue with any system that has more than 32GiB of RAM, especially the typical 192-512GiB RAM IBM X4/X5 servers I've been dealing with since '07. Flushing can totally hang the system for minutes at a time if you leave them at defaults, especially back in the RHEL5 days.

The main problem with Linux has always been it's default tuning, as the same kernel can be used for both low memory footprint (embedded) and clustered supercomputers. Newer Linux distributions, like RHEL6 and, better yet, RHEL7 auto-tune by default (RHEL6's was problematic, and using a static profile is recommended instead, while RHEL7 is much better using dynamic auto-tuning), solving most issues.

So Virtual Memory tuning in servers also rears its ugly head for Virtual Machines too. Platforms like RHEV address much of this by giving one a "canned" HyperVisor/Manager solution (hence the RHEV-H/-M approach) like ESX/vSphere. RHEV-H nodes have a profile specifically set for a HyperVisor implementation, and can also auto-tune (as most of the utilization is often very narrow in focus).

That said, there is also the consideration of read caches too ...

The thing about hugepages is that they cannot be paged out under pressure. That's a plus ... and a negative.

E.g., if you're server is read caching up huge amounts of disk, but you suddenly need to drop caches for a new VM or other memory utilization, it can be very time consuming (scheduler/CPU slice-wise) to drop them, which often involves the TLB.

So ... if you have most of your memory set as hugepages, and you're read caching into most of the remaining, that doesn't leave a lot of memory to page in anything at the drop of a hat. That's why I often set min_free_kbytes at least 10-50x higher than its auto-set value (e.g., 4GiB for 64GiB, 8GiB for 128GiB, etc... -- basically a good 1/16th of memory), to keep that memory from being used for read caching in the first place. You never know when you'll suddenly need a lot of memory, and fast, and you don't want to be dropping caches, and killing the TLB with various syscalls.

With that all said ... and I'll leave this non-Spoiler for "general Linux knowledge" ...

You've read the presentations and have watched the videos of Woodman & Shak's ... correct?

If not, understand every year at Red Hat Summit the duo does a good 2-session (2-hour) tuning presentation. The content is getting more VM-centric and NUMA/cgroup-focused with each year. Most of the information is applicable to any distro, and good, general Linux knowledge, even before looking at KVM-specific tuning info.

Presentations (PDF in 2 Parts) from Summit 2014:
- http://rhsummit.files.wordpress.com...eder-perf-and-tuning-summit14-part1-final.pdf
- http://rhsummit.files.wordpress.com...eder-perf-and-tuning-summit14-part2-final.pdf

Videos (YouTube in 2 Parts) from Summit 2014:
- https://www.youtube.com/watch?v=9BRYrQYc_J4
- https://www.youtube.com/watch?v=qQosbcBvAj0

You might also want to go back through their previous years too, like 2012 and 2010. The 2014 stuff is heavily focused on improvements in RHEL7 (3.0-based), and may skip over some prior knowledge that is little changed from kernel 2.6 releases, or relies on newer facilities in RHEL (e.g., cgroups, systemd, etc...) that all distros might not have or implement the userspace for. Although they still cover a lot of the general Linux tuning basics in even the 2014 release.

Spoiler

Also ... and I'll put this back in a Spoiler block because it's related to Linux Training ...

Even if you don't believe in certification, if you ever have any IT training budget left in your organization, take Red Hat Training RH442. I highly recommend it for even seasoned Linux administrators, and even non-RHEL Linux platforms. If you have a large enough group you can have it taught on-site and save. You can take the class without being a RHCE, you just can't sit the EX442 RHCA (Architect) exam without being a RHCE.

RH442 was originally developed by several of my colleagues, including Paul Morgan who was out at Goldman (and now he's a direct employee with them) while I was at Lehman (now Barclays). Their engineering data centers are next to each other on the Jersey City Exchange Place side of the Hudson across from Manhattan. Paul also, and quite infamously, failed the exam the first time, even though he helped write it.

 

...and my order is in as of this morning! will be a proud owner soon. Unfortunately did not go the 4K option as its the samsung (LTN156FL02-L01) one ;( perhaps sometime next year I'll grab a 4K screen if the price is right and will add more goodies when they become more readily available (more storage, 32gb hyperx impact or vengeance, LTE M.2)

Metabox Prime P650SG Free Shipping in Australia
Windows OS: Windows 8.1 Pro 64-bit with DVD & drivers
Display: 15.6" 3K WQHD Matte IPS Pro (2880x1620) LCD
Processor: i7-4710HQ Quad Core (6MB Cache, 2.5GHz up to 3.50 GHz)
Graphics: Nvidia GeForce GTX 980M 4GB VRAM
RAM: 16GB DDR3 1600Mhz RAM (2 x 8GB)
M.2 mSATA 1: Crucial M500 480GB M.2 SATA3 SSD
WiFi: Intel 7265 AC Dual Band Wireless and Bluetooth
Warranty: 2 Year Platinum Metabox Onsite Pickup

 

Yes, hard drives tend to run significantly cooler, usually less than 40C, while SSD's can get quite toasty even at idle. Keeping the SSD exposed at least should get SOME airflow over it and not tucked between a hard drive and the motherboard.

 

Thank you... I had just assumed that the mSATA would be better until I started reading this thread.

Your advice is much appreciated.

 

I'm still scratching my head on what people are talking about. I've been reading several spec sheets where the same amount of DRAM is included in both M.2 and 2.5" implementations.

Also ... keep in mind that "mSATA" refers to a specific, physical interface (typically the 30x50 Mini-PCIe form-factor with added SATA lines).
So ... it's best to refer to M.2, mSATA and 2.5" as three (3) different, physical interfaces, even if all using the exact same logic for SATA options.

 

mSATA/M.2 SSD Drive
Plextor PXG128M6e 128GB M.2 SSD (upto 770MB/sR | 625MB/sW)


I'm still a little confused.

From what I can gather this has 256mb DDR3.. and a 5 year warranty.

Plextor M6e PX-G128M6e M.2 2280 128GB PCI-Express 2.0 x2 Internal Solid State Drive (SSD) - Internal SSDs - Newegg.com



So why is it a bad option compared to the 250GB Samsung 840 EVO SSD, SATA 6Gb/s (upto 540MB/sR | 520MB/sW) ?

I'm just trying to learn.. I've really no idea which is better.

Thanks

 

Congratulations!

Please don't confuse PLS with Pentile pixel arrangement! PLS is Samsung's version of IPS, and is considered to be quite good. Pentile, on the other hand, is complete crap. The problem with the Samsung 4K screen, and the very good reason not to get one, is not that it's PLS, but that it's pentile.

Sorry for jumping down your throat on this one - it's a bit of a pet peeve, especially considering the panel that we've been recommending as pretty much the best option available (the LTN156HL01) is a Samsung PLS panel, and I don't want people getting confused and avoiding it for the wrong reason.

Just keep in mind that you might also need to buy a new cable - Prema has previously indicated that the cable from a 1080 system will likely not work with a 4k panel.

I hope you enjoy your new system!

 

Yeah they take a different connector as the 4k display requires 4x the data.

 

@dabeer: that was also my initial thought concerning different cable with 1080p screen, but hes actually got himself a 3K screen

 

I'd imagine the 3k and 4k share the same one.

 

Now that we're out of review mode for the P650SE... how's that battery life on every day usage? Someone roughly said about an hour, but wasn't sure if that was the case.

 

Over three hours: http://forum.notebookreview.com/sag...-gtx-970m-htwingnut-s-review.html#batterylife

 

And even the required refresh rate (not just the vertical) inside the panel can be a very, very high overall bandwidth rate, which affects many things.

 

PC Specialist offer the LTN156FL02-L01 panel.. This is listed as PLS. But I thought most people advised against it as the resolution is inflated due to the use of RGBW?

 

Yes, that panel uses RGBW. Yes, that panel is also PLS.

But the fact that it is PLS is not the reason that it is RGBW. These two attributes are not related. You could have RGBW that's not PLS, and you absolutely can have PLS that is not RGBW (thankfully!)

 

Why go with 1TB 5400? Isn't 1TB 7200 provided by default?

 

Good question!

Not form PC Specialist it seems.

https://www.pcspecialist.co.uk/notebooks/defiance-15/

 

It's 7mm vs 9.5mm so you can fit two drives in there.

 

The WD Slim Blue is a 7mm that allows him to have a 2nd 7mm, so he can have SSD + HDD without using M.2.

 

Haha! sniped you, Dabeer!

 

Boom, headshot!

Speaking of, UPS trolled me this weekend... had me thinking my RMA was stuck in PA since Friday AM. I guess they forgot to scan it when it left PA and arrived in CA, because it's apparently on the truck to LPC Digital right now. Thanks for the panic attack, UPS!

 

Question! How are HDDs connected? Directly to MB or via data cable?

 

Directly. You can see the connectors in HTWingNut's pics in his review thread.

 

Thanks! So possibilities of fitting 9.5mm and 7mm drives on SG seem very less to none.

 

On the SE, it's not possible. But on the SG, they have said that they are putting them far enough apart that a 9.5mm and a 7mm will fit.

 

That is nice to know. The official spec of SG should confirm this too I hope.

 

Any new information about the ETA for the SG at pcspecialist or other european resellers..?

Verstuurd vanaf mijn HTC One met Tapatalk

 

Guys how do we know if the screen is IPS?
Looking at buying from bytesatwork.be ( anybody has dealt with them before?) but their description only says ;

• 15.6”(39.62cm) FHD (1920x1080) 16:9 panel, 3.5mm

 

I think we'll need more info (maybe a link?) but based on just that there's no way of knowing whether it's IPS or not - but I'd bet not, because usually if it's IPS the seller makes a point of saying so.

 

They don't say much more. Here's the link to their P651SE

15'' gaming notebook P651SE

 

I also have this question.

 

Price per GB.

 

Email them and ask about the lcd

 

Hello,

Can anyone tell me if any of the external display ports completely bypass the Intel chip (and therefore Optimus). I know many people say this doesn't happen, but I've also heard somewhere (and I forget where and can't find it again) that if you look at the properties of the external monitor in Device Manager, you can see what GPU is driving it. Here is an example on my current machine:



If under "Location" it says it's on the 970M/980M instead of the Intel GPU I believe it means that the connection bypasses Optimus completely. I believe this would lead to much greater compatibility with the Oculus Rift.

If anyone can hook up external displays to the different ports and let me know what it shows, that would be greatly appreciated.

 

Spoiler

Intel's rebranding really bothers me, the Celeron brand was already awful enough, but cramming really subpar Atoms into the group just makes it even worse. Laptop I'm typing from is a Haswell Celeron (2955U), and it's fairly wonderful. Price/performance and power/performance is nearly unheard of, I get about 10 hours of battery life, was only $200, 4GB RAM, 16GB SSD, HSPA+/LTE, awful 768p display. Just something I can carry around with me all the time with no real concern of people stealing, and know it'll last me all day if it has to.

Application is VGA passthru for gaming, Windows guest, swap is entirely disabled on the host. Disk is on LV, yeah. The guest is the only VM on the machine and even when there is no I/O on the host, memory performance in guest is still very subpar, so the read cache issue seems unlikely to me. I may just end up buying more/better quality RAM and seeing if that helps, but I suspect it won't. I'm also forced to use a rather old build of qemu (~9 months old), which might be part of the issue as I know a lot of gaming performance patches have been added, but every time I upgrade, it does nothing but segfault, and no time to really investigate why.

And nope, I generally don't look too much into sysadmin stuff unless it's to solve a problem I'm already having, or it's stuff that massively benefits me and requires a lot of work from userspace to take true advantage of (docker, coreos come to mind). Might check that presentation out if I have time.

I usually am against certs but I do want to pick up some of the RH ones eventually, as they seem to actually be useful and respected (as opposed to the crap like A+). Unfortunately, self employed, and hard to justify the $3,200 for that class.

It is almost definitely a TN panel. It's the cheapest, and if they were upgrading the panel and spending more on it, they would definitely advertise it.

 

I ordered the XMG P505 Pro (based on Clevo P651SE) on 11 November from MySN.

As of 24 November I still don't have it. In fact, it hasn't even shipped yet. Every component in my order was 'in stock' according to their website at the time of purchase.

If you're wanting one of these, be prepared to wait an unconfirmed amount of time. This is really starting to get on my nerves, particularly as I've been provided with 2 false ship dates so far, and now they've not even bothered to respond to my latest support ticket asking for the status of the order.

Hope I'm pleasantly surprised tomorrow with a shipping update.

 

Hey thanks man and all good. Totally understandable, should have rephrased a little better. Just this particular 4K samsung screen is what I wanted to avoid. amended the post to just mention samsung and panel model.

 

did you try to ask them?