*Prospective* News/Discussion: Alienware M18x R3

If this pans out and comes in something smaller, it wont matter what size screen you have.
Brother DumbDumb showed me this.

If it's been posted, then disregard.

 

that mere 10% cpu increment was actually AT LEAST 10% and is now changed to up to 15%, even at 10% a 4.5ghz haswell would be a near 5.0ghz ivy no? there also the Intel Transactional Synchronization Extensions ( TSX) thing ivy and sandy didn't have, although I doubt not many software will take advantage of that but if it did, you're probably looking at at least 30%+ performance increase.

the only rebrand is probably GPU, but its too early to say it now. haswell is not a rebrand of ivy and I think you're mistaken, the tock is always the performance increase over previous generation for intel CPU and a new microarchitecture, something like sandy was average 12% more powerful than it's Predecessor where as ivy was only 5% over sandy.

and no, you probably won't see a hexacore mobile chip anytime soon. at least another die shrink is needed and thats the minimum requirement to fit more cores and be able to cool it off. probably won't happen till skylake or the one after or intel would just go ALL soldered chips instead.

 

We will see. I don't think it will be as good as you so desperately hope it is. I'm not sure why you're so adamant to assume so, it just seems illogical considering all next generation architecture is based of the previous generation, tweaked and fine tuned. I'd say 10% is about right, maybe 15%.

Time will tell. There may not even be an R3, so, keep dreaming. You shouldn't get your hopes up too high. I'm sure it would be very disappointing. It's better to assume 10%, so if it is more, you become even more excited.

And yes, it is safe to assume six core mobile CPU's will be a hot topic towards the end of 2014, 2015. They may not be available, but from what I have read, it's about the right time frame. All speculative of course.

 

Servers will get 12-core/24-thread E5 Xeons at the end of this year, that's almost 3x more processing power with rumored 3.4-3.9Ghz clocks. We do need some 6/8 core mobile CPUs here, otherwise those Clevo P570 owners (if that machine will ever get released) will have a very nice edge in the mobile segment and not even an OC'ed 4930XM will be able to save the day. I wish Dell would make M18xR3 with a desktop CPU and M17xR5 with dual GPU's and mobile Haswell.
Imagine the following specs:

M18xR3: Aka the HIVE!
Modded Intel Server board with support for up to 150W E5 Xeon CPU's (up to 12 cores at least)
Onboard Video for graphics switching -> great battery life
Dual GPU
6-8 sata III/sas3 hard drives (could be some mixture of 7mm 2.5", mSata, etc)
18.4" 100% sRGB 120Hz screen / IPS RGBLED anti-glare
Keep all the good features from current gen and viola!
At least we could have some fun before the world goes completely mobile.

Do I have a free cash to spill on such a machine (top end Xeons are like 2k+) - Hell No!
Would I still buy one if it were released by Dellienware? - Hell Yeah!!!

 

Off topic:
I think if i was gonna go big like that, it would be quad 40 cores / 80 threads.

Have yet to see a 12 core cpu, can you point me to that link? (Not Amd)

 

40 cores would mean quad CPU board and those use E7 Xeons, completely different beasts. Those mobo's can't be put in 18" form factor unfortunately, the smallest I have seen was like 50cm long with no room for a GPU/HDDs/etc. Dual socket Xeon board could probably be trimmed to fit the form factor but it would require 2x the cooling Clevo uses in the P570W -> no go. Here you go, some rumored roadmaps showing an Ivy-EP with up to 12 cores and 30mb cache.

 

Ah, I thought I missed something that was out right now.

 

right now there are E7-88x0 (10c/20t) but those are 2-2.8ghz and 3k+, hehe.

 

Yeah I know, those are the ones Im looking at for a quad cpu setup now. This guys does some serious complex computations and needs the best of the best.

 

Guys, I think I am about to blow my brains out... As I was finished typing my previous question, an IT guy from a company that services the computers in my office came in. He saw that I was trying to buy a new M18X-R2, and he said to me that he "heard" that the M18X-R3 is ready to come out within the next few weeks... I have not read anything like that on the forum. But he also said to me that the NVIDIA GTX 690M recently came out. From what I understood, the 690 came out, not the 690M. So I have some doubts about his words, anybody can correct on my errors?

 

It is doubtful.

I bought an R2 version about a week ago and i did read EVERYTHING on the internet about R3 and every source stated that R3 ( as before ) should come out when newer gen Tech is coming. Which means, when Haswell and Ati 8XXX and Nvidia 7XXX Series come out. AKA Q2 - Jan-July.

If R3 came in the next few weeks a lot of people would be shocked


I hope that helps

 

He's wrong, don't listen to him. Unless he works for Dell in R&D or management, don't take anything he said seriously. I believe the R2 came out at the end of April of 2012. If there is even going to be an R3, it wouldn't be before that. So far, we don't even know if there will be an R3.

Get the R2, you will love it. If there is an R3, look at how much changed from R1 to R2 and you will notice it may not be all that great. In my opinion, the most significant change will be the CPU (Haswell), even the next generation GPU's seem to be a rebrand.

Making major changes involves investing more resources into the product line. Minor changes that make an impact is what will steadily increase revenue. For example, all SATA III instead of a mix between both II and III.

 

If R3 is released, it will probably come out around June, but there could be some leaks around March (CeBit). There's also a slim chance that mobile Haswell will get released around April-May.

 

Not sure why this discussion began as a new thread, seeing that we already have an M18x R3 speculation thread. I'll move it there now.

 

Thanks guys for clarifying things for me. I was starting to doubt myself...



PS: Thanks Mr. Fox for putting this into the correct thread. I didn't think before I posted...

 

word, 10 words

 

b-b-beast, maybe change that to 20 inch LOL.

I dont have the money, but I'd buy one too

EDIT:

70 watt TDP.. thats no easy feat. their server line CPU is much better than desktop/mobile section lol

 

Hey guys, here's another thing I'd love to see fixed/changed: HDMI ports that allow cables to fit without cutting off the edge first. That's one of those annoying things we should see fixed if there's an R3. Adjust the port(s) position by a centimeter, and problems are solved, Dell! This problem may prevent us from being able to use aftermarket wireless HD options.

 

That or cable producers not to add on unnecessary thickness to their cables.

 

I don't think that's a reasonable suggestion. I've owned XPS & Inspiron, used Sager/Clevo & Asus, and none of them had this issue. So, in my opinion, it's Alienware trying to be "unique" with their stylish ports. Not that it's a bad thing, but they could at least sell their own HDMI cable.

That'd be cool. An Alienware branded HDMI cable, ha ha. It has its own lit up Alien Head with customization!

 

Would you like some cable with that price tag J.Dre?

 

The only thing I really wish the R2 had this minute was more Sata III slots, maybe even another HDD bay.

 

You can have 5.5TB storage in your R2, not enough? Also, what would you do with more than 2 sata III ports?

 

I think the problem is that they are using the thick metal chassis that few others use. I think the thickness of the chassis alone is what is preventing most cables from being inserted as far as they need to be. The HDMI port is probably in the same place and the same depth as it is on every other motherboard, but the high quality chassis is what is the difference. They just need to open up the hdmi port in the chassis so it actually sinks into the frame a little bit, instead of resting against it, and I think the world will be a better place.

That and a matte and higher res screen. Then the world would be a better place.

 

msata - 0.5 TB
3 x HDD bays - +3tb
caddy mod - +1tb = 4.5TB unless I've missed something.

I'd like a 4th bay with all of them running at Sata III speeds. Its not much more of a squeeze to have another slot in there as it is now, so I see this as a pretty reasonable request. I also prefer leaving the disc drive in so unless I've overlooked something I don't see how I can manage 5.5TB (which would infact be enough )

 

One correction - optical bay can accomodate a 2TB 15mm WD green. it's a very tight fit but the drive runs very cool, so no problems there. And with 5TB desktop drives scheduled this summer, 2TB 9.5mm drives are also expected (big IMHO).

 

storage isn't the thing for me got externals for that. speed is the thing people want I'd assume more bay = more speed in raid0. also don't forget there are 2TB SSDs, although not out on market for purchase but I think theres a 9.5mm 2TB SSD. or maybe it was 15mm. the LSI 2TB SSD is 15mm not sure about the other one cause it looked pretty slim.

 

I see very little point using external storage with a system with 5 hard drive bays. Only for sequential speeds in Raid 0? Are you willing to sacrifice stability of the system just for that? If we could max out SataIII in every category including random 4k, that would make more sense to me. But even so, what if your raid fails? Rebuilding it would also take quite a while. At this point I'd probably do no more than 2 ssd's in Raid 0, even if I had 10 sata III bays

 

lol 5.5 tb hahaha id do that at 4200rpm speeds truth be told.. i keep way to much stuff on my lappy.. Alienware isnt just for the rich man.. A poor man can scrape and save and make a impulse buy and be happy with it every day after...:thumbsup:

 

Usb 2.0 don't count.

 

Well you made my day. I've wanted that WD green 2tb but didn't know it would fit. Are there any mods needed to make that happen? Has anyone here done it?
For 2TB I can happily part with my DVD drive, then wait for those 2tb 9.5s to come out for some Raid 0 action

 

that 2TB HDD isn't very fast =(. even with 2 black scorpio 750gb, max it goes is 250MB/s and low at near 100MB/s.. slow!! but according to aikimox, it'll fit just fine without any modding and he has done it.

 

I intend to get one of these hard drives when the price goes down a bit. I have a 12.7 mm caddy in my machine and there's bout 2.5mm space above it. Worst case, the I/O board will have to sit at a slight angle of a few degrees, which I don't really care about.

 

guess he doesn't have it just yet lol

speed and performance, with four SSD I just want to see the sequential speed booming lol. with two S301 SLC 4k read goes to 50MB/s, not even two samsung 840 pro 256gb SSD in raid0 can achieve that speed. I wanna see if I can bring that up to 60 with 4 drives

also sequential plays big role in certain areas, it is true 4k are used a lot but it is definitely overrated. 4k write would certainly go towards 1gb/s or above so there are some gain there, while 4k read stay low.


edit: of course stacking sequential only works if integrated hardware raid doesn't bottleneck.. expresscard bandwidth as well as USB 3.0 controller hasnt changed since R1, doubt they would change raid controller. with upcoming full sata 6G, might get to see some bottlenecking with raid controller.

 

we maybe able to see some graphic power jump here.

Real World Labs - 680 Ultra

it is for desktop, but if 680 ultra/titan is released as 780, then maybe 780m would be so much more powerful than 680m.. big maybe, but theres hope to see a 30-40% increase over current generation.

 

There's a whole divided world out there between desktop and mobile GPUs. Believe it or not, AMDs are considered more or less equal to Nvidia in the desktop segment (by some), and driver support for desktop GPUs is optimal. The latest 7990 GPU from AMD is around 10-12% more powerful than the GTX 690 too. I wouldn't compare desktops with mobile's

 

what i meant to say, there maybe a connection. if desktop has a huge jump, theres a good chance we will see a jump in mobile section too. again that is to assume the jumps between desktop from 680 to 780 is similar to 680m to 780m and also assuming 680 ultra/titan is going to be released as 780.

 

You can forget that. We can be really happy if they actually give us a fully unlocked gk114 with somewhat faster memory. The current 680m is choked by it's slow vram.
The 780m should have ram running at 1200mhz opposed to the slow 900mhz in the 680m.

 

whats the difference between vram and memory clock? i dont find vram spec for 680m anywhere

 

more delay? Haswell GT3 graphics to launch in Q3 2013

no extreme mobile processors until Q3, time to go for clevo and it's ivy EP

 

I don't know we may just see a fully unlocked core with low voltage memory still.

 

The memory clocks indicate the speed the vram runs at. 680m 900mhz 115.2gb/s badwidth. If the 780m's vram would run at 1200mhz like the 7970m it would have a bandwidth of 153.5 gb/s which would be a significant improvement.
The 680m gets bested by the 7970m in most games where MSAA is active as shown here: Review Update: Radeon HD 7970M vs. GeForce GTX 680M - NotebookCheck.net Reviews
The slow vram is the only explenation for that.

Sadly that is the most likely thing to happen. I really don't get why nvidia always disappoints with the memory clocks. It's killing their cards.
I'd rather see a fully unlocked core with slightly lower clocks but significantly higher clocked ram. We can always oc the core with bios mods.

 

Faster GDDR5 takes a lot of power, by lowering that they are capping the highest overclock but on balance at stock likely increasing speeds due to a higher core clock.

 

The 680m is clearly running into a memory bottleneck with MSAA. Nvidia should resolve this problem with their next gpu. I can bring a nice example with BF3.
The 680m has 60% more raw shading power than the 675mx,stock vs stock,but it only out performs it by 25% when all settings are at ultra(based on notebookchecks results).
They have the same memory bandwidth so the 680m is obviously running into a memory bottleneck. The 780m would deffinitely benefit from faster ram.
If they unlock the core but leave the memory at the same clocks than the 780m won't be much of an upgrade. It'll still be running into the same bandwidth wall as the 680m.

I do wonder though. Is it possible to overvolt the vram through the vbios?

 

No and I had a look at the power circuitry and it's not clear how to hard mod it either.

 

Disappointing.

If AMD can squize 1200mhz ram into their top mobile card than Nvidia with their more efficient Kepler arch should also be able to.
Otherwise they'll be bested again by the 8970m when it comes to anti aliasing.

Anyway. Nice chating with you Meaker. We'll see how things play out in the summer. Maybe sooner.

 

theres a chance for 8990m btw but until that chart is fully released, nothings confirmed

 

Why would it be called 8990m? It's gonna be named 8970m and I'm sure it will have great hardware specs but bad driver support and far less OC potential compared to the 780m. As usual from AMD.

 

The 7970m uses a mid range chip in a different class to the gk104.

 

Yet it out performs nvidias top dog. There's no excuse for nvidia always going with low memory bandwidth and putting most of the tdp into the core.
It deffinitely improves 3dmark11 p scores since there's no AA involved but games are another story.
7970m also gets a better vantage score because 3dmark vantage is also very bandwith hungry.

Would be nice to see a 384bit mobile gpu with Maxwell. They could use low clocked ram and still get a high bandwidth.
I know,I know. It's not gonna happen.