*** Official Clevo P870KM1/P870KM1-G/Sager NP9876 Owner's Lounge! - Phoenix 3.0 ***

Those seem like great temps - you show off HAHAH j/k @Mobius 1

I have now used the double heatsink as well, and OMFG this thing is a godsend, and it's HUGE...it also weighs INSANE number of lbs extra.

Was better balanced in terms of making contact as well. but the laptop was literally 20lbs with the heatsink and SLI and drives etc...

20 friggin pounds for a 18" laptop. not 19 and change...TWENTY....

 

you sir, have to play with your fan settings until you get it right. If you haven;t checked contact with the CPU and GPU to the heatsinks either and repasted or delidded - then you need to maybe get that done too

 

this is not hot, actually this is decent - because not only is your trick with the under volt offset working, your not throttling at all, AND it shows 100% usage which is friggin INSANE - like do not bench at 100% your just asking to fry something for no gain ! Like so a lighter bench like XTU or something, **** you might as well be running Prime95 which I do not recommend.

 

Yeah this is happening to many people.

Not really sure what the update broke, I think it might have re-installed over existing profiles and settings for drivers, like your NVidia drivers and such...

same thing happened last year with the anniversary update

 

is my temp actually good or?

 

I will give you my best advice for this... go down to 1 card. Physically remove the 2nd card.Cable too. Remove the driver and uninstall it and delete if in safe mode, reboot. It will auto-install or have the Microsoft basic adapter. You might have to disconnect from the internet before all this, just to see that it goes to the basic adapter. If not, rinse repeat in deleting and uninstalling at same time while not connected until theres no driver for it but the microsoftbasic adapter.

Then install the latest driver. Reboot, check it/test it, configure it use it etc..

once it's verified working, shutdown, install second card and SLI cable, and start it up, without internet and install the driver again, yourself, manually. Itr should now enable SLI and be fine, your settings you had in the control panel should all carry over as were good in testing, and you should be able to hit SLI after reboot and it should work.

If the latest driver doesn't work, try the latest WHQL driver posted on your laptops manufacturers site.

 

thats about to be expected.

 

Eurocom had a special or might still, on the 780W PSU's, I have one so I don;t have to try different PSU's or open up the packaging from laptops intended for you guys when you purchase from me, and it eliminates me having any issues bottleneck wise due to power. It's loud, more then I am comfortable with, but I use it at my bench, where I work, not for powering a laptop I game on, the X9E3 I have really does make it whine out and make noise though - but it works great !

 

Theres a thing where if it's not under X% it won;t charge until it needs too. This is a good thing, for conditioning the battery, however there is a setting in the bios you can set that will make it always stay charged and charge for almost no reason sometimes. This might kill your battery prematurely though, like 1-2 years earlier.

 

Pheonix you Mastersmith ! Good tip ! 5 rep for you ! (cuz you need it riiight ?)

 

Charging it 100% is bad advice, unless you take the battery and stick it in the freezer once it hits 99% otherwise that extra heat will oxidize it and lower it's potential, if not melt the insulator between the chemicals inside and maybe even explode one day !

 

what ar eyou talking about lol, right above the video it says optane as OS drive, so entire OS is on optane 32gb cache drive.

yep, with primo thats about the performance you get though if your software doesnt take full advantage it is pointless. like i said, ram disk has its use for exmaple putting my entire browser profile on it make it more snappy than being on SSD but not that much faster. but optane is great for booting up faster and much faster in 4k than NVMe pcie SSDs, something about 6x-7x. if you don't see a benefit, you don't have to get it LOL. on the other hand, I'd rather have single optane storage device than two NVME PCIE SSDs in raid 0 simply because in mix workload, regular NAND flash gets destroyed and performance drops a ton where as optane stays fairly consistent.

also ramdisk has its own share of issue too, in my example entire browser profile into ram disk. not only browser is memory intensive on its own, it also need to use partial performance for the profile, result in stuttering when ram is underload. if I wish to do video editing or make heavy encoding/compression and needs a lot of ram bandwidth, while running browser i'd get stutter everywhere lol. theres pro and cons, having optane to take a load off of ram would be awesome.

 

After delidding and a sligth usual OC, you should see 50C at idle 55C while browsing and fans barely moving, all the way up to like 75% fans speeds while gaming for long sessions and temps at about 85C thats about the sweet spot for normal 25C room temps. Hope that helps. If your hotter, then you can make easy improvements, if your lower, then your ahead of the average user.

 

We updated the driver page to have the right drivers now, should be fixed, let me know if it's not (on Eurocom site)

 

what are you looking for, I have extra's of all the equipment for this laptop ( had to test this stuff out for myself)

 

I might..let me look....



damnit all I got was before I put it on....

 

That is correct, the clevo 1070 is like the 980DT

 

Yeah the laptop was released a bit early, like the F5 was released before it was perfect too, this was done to benifit those who "just can;t wait" to have the latest tech etc.. and hopefully you crazy people will help them figure it all out.

(I knew I was crazy to buy this laptop....lol)

 

I always use a ramdisk for my browser cache, thought I was the only one... Like when i load up Chrome I'm loading up 12 pages and logging into various sites, and because it take sup my whole screen too, I just close the entire browser like 100+ times a day so that extra boost in speed is HUGE productivity increase !

If you have a HDD that comes with a Cache drive thats huge and does this, thats amazing - but how does it get speeds to match RAM without using the systems RAM ? Otherwise it would be stuck at NVMe speeds at best - or does it use the systems RAM, like does it boot to the HDD then loads into system ram as a partition to boot windows off the RAM ?

 

we wont really know. caching is a thing that needs to be taken advantage of, heavily, and software has to be coded/optimized for it or smart enough to use it and if they dont, having a cache wont do a thing and remain slow as HDD. it's because of just that uncertainty, i dont like cache in general because I don't understand the software itself. in our case we put entire browser profile into ram so we know for sure everything is running as fast as possible and not having to worry about it know 100% its being used lol, even if chrome to not take advantage of ram as cache per se, it'll still be running at high speed cause the profile is on ram disk.

which goes with the reasoning that I will get optane for default boot device, not caching device imho thats just silly cause it fully dependent on the software. now, if i were able to code it and recompile it that would be a different story.

 

Meaker did say this, and to be quite honest, I agree.

But there's a difference between "unlocked" (see: @Mr. Fox pulling almost 900W in a benchmark) and "enough for stock" (see: the opposite of what this system has). I'll say it again, even if the power limit for the thing was 600W, it'd have been fine. 190W + 190W + 95W = 475W for maximum stock CPU and GPU loads, + an allowance of about 125W for the rest of things a system could be doing (charging a phone, running USB hub with some fans, etc).

480W is crippling stock, far less what someone who wants the most out of a system wants.

Edit: if I may add something else... I also seem to remember at some point Meaker saying he had to be careful with a single brick, even using single GPU, because of the unlocked power draw. Should this arguement be made at all, I must make the counter-statement of: "This machine does not officially support single PSU."

At least, the last I heard of it, starting with the DM3 version.

 

What do you mean "What are you talking about?". I just gave you a print screen and link to the page. It clearly says "Optane Memory running as our OS Disk on TweakTown's Z270 SSD testing system". This was the system that achieved the 10.5 second boot time. The other video's title said: "1TB WD Black mechanical HDD + Optane Memory on the system provided by Intel" which had a 14 second boot time. Nowhere did they show a video of just a pure SSD, without Optane, to compare the boot times. This has been my entire point that you've conveniently ignored. The irony is, you keep telling me to read, when you yourself refuse to do so.

Also, running raid 0 NVMe drives has to be the most pointless raid configuration to exist. You are going to be severely handicapped by the DMI bus, that you won't be exceeding the raw sequential speeds of a single drive by much at all. It may help in random 4k, but even then, it's a serious waste of cash for arguably very little benefit. As for boot speeds, you must not have an SSD. We already have boot speeds that are as fast as Optane, on SATA SSD's. My 850 Evo, as old as it is, can still boot in 10 seconds with my fast boot settings (and no startup applications). You honestly expect Optane to shave off a noticeable amount of time? You are in for a world of disappointment if that's the case. Also, you've yet to post a single drive that is even remotely close to my ram speeds as a cache.

You also keep saying "if your software doesn't take full advantage". Primocache is software agnostic. It does not have a single compatibility issue with software of any kind. It's a block-level cache. As long as you have enough ram to get the job done, it will do exactly as you configure it to do, for literally anything on your system. Any program that does block-level caching can do the same thing, it's not limited to Primocache.

The best configuration, is still going to be: large spinner, large SSD (to cache spinner with level 2 block-level cache) and copious amounts of ram to defer mundane writes to the SSD (level 1 block-level cache with write defer enabled) on a system that has a backup battery. Bonus points if the drives themselves are in raid 1, for extra redundancy on top of the performance boost. No RAID 0 in the world can currently top the speeds of ram cache anyways, so your spinners will feel plenty fast as is.

 

Problem might be that the 4-pin DIN port doesn't have a sense line. So it can't tell the machine what the charger is rated at.


More of a safety feature that allows the system EC to not overdraw from the PSU, but mainly abused by manufacturers as a DRM to not be able to use other manufacturer's charger. IE: The Omen 17 and Alienware have the same charging port and charger brick, but the Omen 17 won't accept the Dell 240w slim charger and vice versa.

 

Caching has been the thing the industry has been moving towards since consumers found out servers did it in the 90's.... it's just that great of a feature. Essentially an NVMe drive is like a massive caching drive.

 

Whats the maximum average throughput or bandwidth we can expect from a NVMe, since they run off minimum PCI2 2.0 x 4 speeds ?

 

Unless or until power draw exceeds the peak output capacity (not the nominal 660W rating) of a dual 330W AC adapter setup, there is no legitimate rationale for "protection" to be present. If a legitimate need for that exists, the explanation would be that Clevo erred in the engineering of the product and released it with defects instead of fixing their mess. I doubt that has as much to do with it as incompetence and stupidity. Otherwise, the explanation would be that they are purposefully and maliciously marketing a known defective product. I do not believe the latter, unless they leave it that way instead of taking swift action to correct their firmware engineering blunder.

 

NVMe on PCI/e 2.0 x4 cannot sustain over 2GB/s in either direction. PCI/e 3.0 x4 is 4GB/s.

Sent from my OnePlus 1 using a coconut

 

But some NVMe is running on 2.0 I take it the faster is running on 3.0 then.... but wow, we're already at the throughput of the next gen standard that just came out ... darn.

 

ummm u gotta know that not all software take good advantage of ram let along cache. if you put say the temp folder or work folder into ramdisk then yeah it'll take advantage of it cause you know files are being written into it. there are even software better off with single rank ram rather than dual rank etc

 

yea but enterprise environment is mostly about ram since forever lol their program are all optimized for it.

 

There are no "temp folders" or "work folders". Go Google how block-level cache works . Clearly you've not been reading anything I tell you, which implies you either don't trust me, or you simply don't believe what I say. Plenty of other sources out there to corroborate my word.

 

i just clal my entire ram disk temp folder/work folder, i think you are assuming a bunch of things lol.

 

You must have yours configured entirely different than mine. Mine is not a folder, and it's not seen as a "disk" of any kind to Windows. It's literally a block-level cache that caches anything I use. Completely software agnostic, and works with anything. Yes, it doesn't help my boot times, but when I am already booting in roughly 10 seconds, I don't need it to. If you have enough spare ram, I highly recommend giving Primocache a try. They have a 60 day free-trial that has every feature as the full program. I can even help you set it up. While I refer to it as a ramdisk, it's fundamentally different. Yes, it uses ram, but it's not exactly a disk. It's extremely fast cache, that is lower-level than even Windows Superfetch. Superfetch will pre-load your DLL's and exe's as it functions on a file-level. Primocache functions on a block-level, meaning that it doesn't have to cache the entire file, just the specific blocks you use the most.

Again, I wouldn't do this if your system isn't 100% rock solid, as any crash of any kind can require a windows re-installation. I only advise using level 1 ram cache with write-defer turned on if you backup often and have your PC plugged into a UPS.

 

thats prob where the difference is, iirc in primo theres option u can choose to have it a usable disk or not. mine and woodzack is an actual disk, and imo regardless which option u choose it should only increase boot time, not shorten it as it needs to load it prior to window properly starts, especially with all my temp files and junk are placed in there. superfetch is written onto ssd and loaded into ram at times for faster access.

now, how would you make sure what you are doing are fully taking advantage of ram, we can create ram disk for caching, but how much does our daily window/software uses it is another different story. a ramdisk won't be as fast because its managed by software so performance dropped by at least 1/5th shown by pcper, though at that difference we can't really feel it.

 

Primocache has a feature built in to tell you how often it is utilized. It's called "Cache Hit Rate". The higher your cache hit rate, the more things that are being cached and used.

When I tell you it's software agnostic, I meant it, lol. 99.45% of the blocks I access, are being cached. Look at the cached read. Almost 2TB worth of blocks that were read, 870GB written, with a simple 16GB ram-cache configuration. If you had an even larger allocation of memory, you would likely do even better. BTW, that configuration is with 4kb blocks. Also, it has not increased my boot time at all. If needed, I can take a video of my boot times with it on and off if it will help. As a matter of fact, if you use the level 2 cache (SSD to cache HDD), it will actually improve your boot time drastically. That was the preferred configuration I mentioned earlier (level 1 ram > SSD level 2 cache > giant spinners). Hopefully when I pick one of these laptops up, I can show off it's high ram capacity and it's many storage solutions. Might surprise people how well these things will do as a tiny little file server, lol.

 

i took a look at it and tbh it doesnt explain a whole lot. i self taught and often i waste time on stupid little things. 99.45% is a number and imo sounds very satisfying, but in the end its just a number. 99.45% of what files are being hit, are being used in cache? say a software access 10 different files, only 1 is being thrown into it for cahcing and that 1 file being hit 99.45% then its great, for that single file access, rest are pointless no? (assuming they are accessed at a comparable rate) software optimization is what im after, which i'll prob never be able to figure it out as im not a programmer and even if i was, i wont have time to re construct entire software to take full advantage of hardware ie ram/cpu extension etc.

on the other hand me and woodzack was saying is that, the example i used was throw entire browser profile into ram. since i dont understand the aspect of what is being used and will be used for cache purpose, putting entire thing in there would target that problem for the most part, as i know for sure anything is being accessed by the software from my profile will be loaded from ram, but of course at a slower performance.

 

That's the problem. You keep thinking this is about files, when it's not. I keep using the word "block" for a reason. I don't need to cache an entire file in order to get faster access to it. I simply need to cache the parts of that file that I use the most. That's the beauty of block-level caching. Rather than trying to stuff an entire file or folder into a finite amount of ram, I only need enough ram to cache the parts of the file/folders that I access the most. It's also why I have an extremely high hit rate. It means I am caching the vast majority of the things I use, thus resulting in faster performance (and less physical wear on my SSD's due to deferred writes hitting my ram instead) because I am accessing them again from the ram cache, and not the disk itself.

If you would like the definitions to the data in that screenshot, here they are: http://www.romexsoftware.com/en-us/primo-cache/terms-performance.html

The reason why Optane (in it's current state) seems unimpressive to me, is because it's SRT on steroids, for very limited scenarios. You can cache an HDD with an SSD, just as effectively as you can with Optane. Need specific queue depths? Buy an SSD rated highly for the specific queue depths you need higher performance for. The vast majority of people that are buying new platforms, tend to be doing so with traditional SSD's, and Optane simply isn't helping traditional SSD's at all. If it is, it's certainly not to the same degree as Optane helps HDD's, which begs the question, why spend the money on Optane in the first place? When 3D Xpoint hit's much larger capacities, it might become interesting for different reasons, but from a strictly cache-based perspective, there are simply better alternatives that work for more than just their current platform.

 

let me ask you, why would block be any different?
also for optane, 32gb is junk, using it to cache HDD is onething but I want it as my boot drive that would overall speed up everything that I have installed on it. limited use or not its based on the size and how you use it i guess

 

I've explained this several times over. File-level can only cache specific files (DLL's, exe's, etc) while block-level can cache specific pieces of each file that is accessed the most. It's more efficient, thus, faster. You seem to be doing zero research on your own part, and I question why that is. You also seem to be misunderstanding what Optane is in it's current state. It's not a boot drive. It's a caching drive, extremely similar to Intel's implementation of SRT. I've presented a ton of information for you to digest, but you keep ignoring it and asking me the same questions over and over. Please, take the time to read what I've presented to you, and study up on it. It should make sense in and of itself. I am also certain that a quick google search will yield the differences between block and file level cache systems. You can also test this yourself with Primocache's free trial, and compare it against the ramdisk you are already claiming to be using. You can even post your results as evidence to use with your own claims, if it happens to support what you are saying. Just understand, memory is far more complex than simple frequency speeds. Rank interleaving (the ability to read from one rank while writing to another) has a drastic impact on ramdisk/caching performance. The tertiary timings that are typically associated with Rank Interleaving (most commonly listed as _DR suffix) also impact your performance with these configurations. tRFC and tREFI also has great yields in latency as well, so all of this will impact your results if you leave yours extremely loose.

Part of the reason why I hope Premamod bios gives us a little bit of control over the memory sub-timings.

 

i think you simply misunderstood what i said so let me clarify somethings, i know what you're saying "File-level can only cache specific files (DLL's, exe's, etc) while block-level can cache specific pieces of each file that is accessed the most. It's more efficient, thus, faster"

though that goes back to what i mentioned b4, percentage is a thing, too bad when i mention file and block, in my head i call them the same. u see since i self taught, my learning is incomplete. i dont use technical terms that people uses and they may have a specific name for a specific level, as long as it make sense in my head then i understand it lol and thats probably where the issue is.

now in this case, i now understand theres a file level and block level and you are trying to separate them. ultimately, they are still data on disk. data caching or not, thats window managed sometimes software managed, to take advantage of hardware, it changes nothing from what ive mentioned.

theres always uncertainty of what data are being cached, i throw entire "file" into ram disk and solves my problem. unless you can somehow prove and convince me that windows somehow knows what files are accessed the most and caches all of them, which im sure window does to an extent but not all data. in the case where not all data are cached even most used ones, optane will take care of that. aside from that, i'll throw all my most used read/write files into ram disk without crashing my windows and rest will be optane and from that point on, window and other software can cache w/e they need to cache.

theres the optane benefit, you may call it niche or what not but thats how i intend on using it. 128gb minimum and perferably 6 to 7 chanel would be fastest. thats way better than having to have 48GB of ramdisk for just caching.

 

It's not windows managed at all, unless you are using Superfetch. That's the entire point of primocache. It's a layer that is in between windows, and the I/O coming from your drives. Think of it as a net. When you first catch I/O coming from the drives, it passes through primocache's net, then goes to windows. Instead of Windows getting the same data from the disk again, it will instead grab it from the net, bypassing the need to grab it from the disk, saving a ton of time. If you have write defer turned on, it works the same in reverse. Instead of Windows writing to your disks, it will write to your Primocache first. Once the primocache you created is full, it then dumps the data on to the disk itself starting with the oldest-cached blocks. It's completely seamless, and you never even notice it happening. It's also completely agnostic to all software. Literally everything you do, any file/folder you access, get's cached on a block level. You can even configure it to intelligently monitor which blocks are accessed the most, and have Primocache save them on shutdown (level 2 configuration) which will speed up boot time and access times to the same blocks.

You keep speaking in absolutes about Optane, as if it's somehow doing something Primocache isn't, but I've already explained it to you. Optane still uses a similar configuration as Intel's previous attempt at caching, SRT (Smart Response Technology). https://en.wikipedia.org/wiki/Smart_Response_Technology. This manner of caching isn't superior in any way to how Primocache handles the caching. You also keep mentioning that you intend to put your entire OS install on the Optane accellerator, but you've failed to research the fact that These small Optane "sticks" don't work like that. It's an accelerator, not it's own drive that you can just throw windows on. The RST driver requires another drive to point to, in order for Optane to function. http://www.intel.com/content/www/us/en/support/memory-and-storage/intel-optane-memory/000023853.html

"Note that Intel ® Optane™ memory accelerates the files and applications that you use the most by monitoring your usage for optimal performance". It's literally doing what Primocache does, but slower (due to the fact that it's raw throughput and latency is weaker than ram). I've yet to see a single reviewer give it praise over a normal SSD, other than it's extremely low latency at lower QD's. Again, ram can do that too. Here is a "consolidated review" that covers Arstechnica's, Anandtech's and PCper's review of Optane: https://www.theregister.co.uk/2017/04/26/burst_of_optane_memory_reviews/

You seem to think Optane is going to somehow be better than traditional SSD's, and still be cheaper. Look at it's current price:GB, and tell me you expect the price to be even remotely competitive once larger versions come out. Intel will always charge a premium, because they know they can. Part of their strong brand recognition allows them to do so. Whether or not their products are always the best choice, is a different story. In this case, I'd have to say you are better off with other SSD's, even when using them to cache slower drives.

 

There is a difference yes Though 125W for the rest of the system is total overkill for a heavy gaming scenario

It's something I intend to raise.

 

Yeah, I know it is, though the system seems to draw a lot. The two cards and cpu should be 475W max but as @Mobius 1 found out with his CPU being limited to 55W under full gpu tilt, the rest of the system needs 45W+ and this is with stuff turned off. So I would say 550W should be sufficient

Sent from my OnePlus 1 using a coconut

 

Can Sager owners please put together their figures showing throttling at stock and pass it on to support, we can pass this back to Clevo and get it looked into

 

card 2 throttling on gpu only test

will get cpu later

 

so what you're saying is, primo the software see these IOs and intercepts them however you call them and then cache them and its not window managed. if its window its likely driver/services to manage it or how its designed but primo is a software, in the end of the day a software could be better made than window's default but still be flawed.

besides, like i mentioned even though its showing you all those data, there still wayyy too much unknown stuff to me and i'd like to rid of all of that uncertainty and go for something i can easily have control of and know for sure it'll work better, that is to get an optane SSD as boot drive with at least 128GB in size and 5-6 channel of flash die.

since we don't know what the parameters and algorithm primo uses, u'd never know how it treats IOs. if u knew that it works well, i sure would love you tell to me the set of rules it follows, such as data IO being cached after load first time, 2nd time, 3rd time etc, how long that data stays and being treated as cache data before its removed. how big the data is and what type it is, sequential, random read/writes, when is it removed and how the data no longer being seen as needed for caching etc etc.

even if i have the above facts i'd still be unsure as i have no way to test it out, what you mentioned to me about block data io etc etc 99% etc those are just numbers and i dont see them as any advantage to me. though one thing i do know for sure, is that if i replace NVMe PCIE SSD with optane storage, my default 4k shoots up 5-6x the performance at QD1, so with just that it'll already be faster than your NVMe PCIE SSD + primo cache, because I can simply do the same, optane storage and primo caching, or ramdisk. btw, those "block" of data you mentioned, yea they are mostly random reads, thats what caching is for, and first time it loads from storage device, optane 4k random read is 5-6x of PCIE SSD.

i donno if you're simply blindly following articles/guide out there but doesnt appear to me that you have thought this through.. even with explanation, theres always the real world use and a lot of times turn out to be different than expected..

 

They are not going to fix that i'm afraid. That is gpu boost working against your temps of 67C+

 

Please use a combined test to increase the combined draw and ensure it's a board throttle you are talking about.

 

I honestly cannot tell if you are trolling at this point, or if you are being intentionally ignorant. I've told you several times that they have a 60 day trial period of their software, with every single feature as the full version, for you to test it out. You keep claiming you have no way to test it, but that's simply not true. You also say I am the one "blindly following articles/guides" but you don't even have Optane. You keep saying you will use it as a boot drive, but Optane, in it's current state (unless you are referring to the P4800X, the $3000+ SSD) CANNOT be used as a boot drive. The only person that seems to be lacking a fundamental understanding of how this works, is you. You've admitted this several times. Even as I present mountains of evidence, you ignore them blindly, and repeat the same nonsensical words you've been saying since the start. If you could provide evidence of Optane being used as a boot drive by itself (and not as a cache drive for another drive) then I'll gladly admit I am wrong. As of now, Optane is simply a cache accelerator, and a niche device in an already niche market.

https://www.romexsoftware.com/en-us/primo-cache/faq.html
"
Q3: Can I restrict cache to certain folders or files, or can I exclude certain files from cache?

A: No, you cannot. PrimoCache is a block-level caching program, residing below Windows file system drivers. It cannot capture correct file information." Optane uses a very similar method as well, if you bothered to even watch that video (from Intel themselves mind you) detailing how it works. It's loaded before Windows file system drivers as well. Like Primocache, Optane only gets faster as you repeatedly use it (something you can easily see by watching any of the Optane review videos). At this point, it's clear to me that you've done little to no research on Optane, or caching methods of any kind. You clearly expect Optane to be something that it is not, and worse, you expect it to be a solution to a problem that doesn't exist. If you'd like to continue this discussion even further, let's take it to private. We've eaten enough of this thread as is.