**Official Sager NP9270 / Clevo P270WM Owners Lounge**

Regarding the real-world benches, I did some testing of ssds in SATA3.0 vs SATA2.0 and there was no noticeable difference other than in synthetics. I also did a RAID SATA3.0 vs single and again, there was very little gain, only really seen in benches. And with RAID0 you still have a bottleneck in 4k benches. Plus, you are paying extra for the extra SSD. Plus, you are sacrificing the storage. But be my guest, wait till Haswell, go for 4 ssd's and report back.

 

from my experience, i cant stand raid 5. having to rebuild its array incase something happens, way too slow. raid 10 is much better, or just seperate two sets of raid 0.

not sure if most of you know this but any SSD you see nowadays on most review sites DO NOT have their OS installed on the drive itself. the best tool to test window regular usage would be HD tune because it tests SSD for high compressible file at QD 1. test the drive when its empty in another computer and test it again when you have files on it and you'll notice massive performance drop, something around 30-40% (for the area/partition where files existed). for example if you have intel 520 240gb SSD and on review sites it shows 420mb/s read, then when theres file on the drive you'll get around 300-350mb/s and thats the actual speed you'll get for QD 1. again, SSD dont work like regular HDD, HDD is fastest when its storing files on its most outer side of the platter.

thats why sata6gbps and sata3gbps dont have much difference simply because with the same SSD, you're really looking at something 350 comparing to just below 250-300. however this is different for sequential incompressible data which is mostly for movies/music, zipped files.

having 3 SSD in raid0 with OS installed and running on it would finally make it 1gb/s read at QD1, and has to be some of the best SSDs to be able to accomplish it.

 

I understand you and you do make a lot of sense. The problem is that you don't necessary use the machine for anything particulary important to be able to notice a difference.

The difference is impossible NOT to notice (for me that is) dealing with virtual servers and shared servers and a lot of files opened simultaneously one does notice the difference. But hey, not everyone does that.. a small percentage does, I guess that's why they need a machine like this

 

i never have had raid0 SSD experience and i am desperate =/ ive got raid0 for HDD for its transfer speed which is awesome though. definitely need that haswell cpu/chipset and hope for a thunderbolt lol.

 

btw, did anyone get annoyed at the idea of having expresscard slot at the front of the laptop? i kinda dislike it it would jsut get in the way of using keyboard. im not sure if this will be carried over to 570 but that was the old model. should be much different from the new one.

 

The transaction from RAID 0 HDD to SSD is insane, you don't need RAID 0 SSD.

 

i think you misunderstood it. SSD for storage is a bad idea unless im using SLC, MLC drive will die considering im copying around 100-200gb of files a day. so i have no choice but to use 2 HDD in raid 0 which write cycle is suppose to be 600k which is 6x more than SLC but again HDD can die of other reasons fairly easily. Also HDDx2 gives me 1.5tb rather than 240gb.

another thing is no point using SSD internal storage as of now because the max speed i ever get out of my alienware is esata 3gbps which is under 300MB/s. usb 3 isnt good at all and i dont have an external device for something over 6gbps.

 

That's exactly what I do, lol. I have a R2 + Hyper-V with a forest of domains on a SATA2 SSD and am yet to see any bottlenecks there. I can reboot 5 DCs (loaded with roles and software) simultaneously in 5 sec. I have an Exchange 2010 DAG loaded with Jettest and Loadgen to simulate a real-time 1000+ user mail (2TB/sec) traffic across the domain, + taxed SQL + BackupExec server + remote agents. And I still don't see any difference between a single SSD and a RAID0. Why? Because in real time, sequential read and write is only a small portion of a huge spectrum... But I do see a shortage of disk space, some simulations require many TB's, and resorting to external arrays ruins the mobility.

 

i think most used are 4k read so sequential read dont mean much unless its file transfer. and too bad 4k read doesnt stack up like sequential reads, more like stacks by 5% lol, 10% if really lucky with some massive tweaking in raid config.

 

Point taken guys. Thanks for the replies ^^

 

clevo/sagers, make a 18 inch laptop =)

 

They used to. For this current generation, Sager/Clevo is only offering a 17.3-inch laptop even with dual graphics cards. This may be subject to change, but I do not see this changing any time soon.

 

=( good bye to 18-20 inch laptop for good?

i remember clevo was one of the company made a 20 inch back in the days. with acer, hp, dell, alienware and clevo`s 20 inch had dual graphics.

 

Getting a 20" is ridiclious now a days as machines, workstations and screens are everywhere.

But we should see a 18" on the play soon.. as Heihachi mentioned so clearly.

Soon being in between the release date of HL3 and Blizzards new Diablo.

 

A bit off topic, but FWIW, some posters are now reporting resellers are telling them the P370EM w/ single AMD 7970m will ship with Enduro, and dual 7970Ms in XFire will not.

Not sure what actual truth is on this one.

 

That should not be possible... did a reseller report this or just normal posters?

 

i know that for a fact 680m single card will OC close to performance of stock 680m SLI, but 680m SLI can only OC about 130mhz each before throttle happens.

so im not that surprised.

also, it seems S301 SLC drives will finally be out on the market soon.. maybe next week or within this month.

 

It's all hearsay at this point. So take the following posts w/ a grain of salt.

First the poster says a reseller told him it did have Enduro for single card -
http://forum.notebookreview.com/sager-clevo/687130-intending-buy-np9370-3.html#post8839183

and

then tries to confirm with Sager where an email says one thing and a phone call another - http://forum.notebookreview.com/sager-clevo/687130-intending-buy-np9370-4.html#post8840286


My guess is it does not, but would like official confirmation from a reseller or builder that is indeed the case.

 

Yea the outputs are wired to the dGPU's, so Optimus/Enduro a little impossible. Also on the official Clevo model description and documentation there isn't a single mention of Intel Graphics, nor do they provide reference Intel Graphics drivers for the P370.. So yea MYTH busted.

 

+1 rep for the pun.

 

Trust me we work VERY hard to come up with them

 

Back on topic: We're midway through September, and yet there is still no new info on the P570WM? All the rage and focus has shifted to the P370EM.

While the P370 looks like a great machine, it's no desktop X79! I still have my heart set on the P570WM, so I've been waiting ever so patiently for new information. Can any resellers share some info, or maybe bug Clevo for some tidbits? Mainly if the release is still on track for late November to early December, or if anything from that June specsheet has changed (like inputs).

 

Where did this come from?

 

heh, techinferno forum. they got great individuals there and people who mod vbios etc, or bios in general. as for this particular discussion for 680m SLI nVidia GTX 680M Performance Review

 

They didn't mention temp. I wonder what it would be with oc'd 680M SLI.

 

um anything new? its almost october lol

 

Wake me up when November ends.

 

My guess is it's gonna be after Xmas.... By then, the current chassis design will be so obsolete, hope they redesign everything from scratch while at it and make the system thinner and of better build quality.

 

maybe you didnt see this thread. Alienware M18x-R1 + 680M SLI

 

It's so odd that Nvidia won't drop the 680 price after seeing its performance is so in-line with the 7970...like really, $500 more than the 7970 on dell.com? for WHAT! a good portion of us won't be OC'ing it....

 

I don't think so. My hunch is the number of units sold for the 680M is meeting their expectations.

Performance is irrelevant at this point. Until sold units comes in under their estimated numbers, there won't be any drop in price. However, if more and more consumers see these performance numbers and don't see the benefit in paying the "nVidia" premium, and unit sales drop as people move to the AMD 7970 alternative, then they may be inclined to make a move. Until that happens OR new tech comes out, you can expect to see the pricing of the 680M hold steady.

 

http://forum.notebookreview.com/alienware/674627-nvidia-680m-not-overpriced.html

Guess that pretty much sums it up, love the comments on comparing apple to nvidia

 

Yeah explains why they are hesitating with releasing it, delaying the deadline

 

well, for desktop nvidia and amd graphics are similar in cost and performance, only mobile graphics are far apart.

 

so they rumored ivy E will have up to 12 cores while sandy E took 6 active cores of the 8, would ivy E have a chance to go to 8 active cores? 8c/16t or still stick with 6c/12t

 

Well the die is capable of holding 12 cores, which is pretty beast. xD
The number Intel chooses to "enable" is up to speculation; I don't see what other reason they would release IB-E after Haswell if IB-E didn't have 8+ cores.

One thing about this laptop in general, Eurocom will be releasing a Panther 4.1 in mid/end of October with X79! =D I emailed them [regarding the P270WMr] and the reply I recived:

Will any other companies also be using this Clevo body?

 

That would be nice, if only Eurocom is more trustworthy.

 

I would not get my hopes up from anything that comes from Eurocom.

The only thing we can go by for their shipping dates is that it means it's usually up for pre-order from Sager 1 month after Eurocom says it's supposed to ship (it never does).

So I guess that late November guess was right. Cool. Looking forward to it!

 

the above, and x79 isnt my thing. quad channel memory is something to brag about but barely any performance gain. Clevo holds less harddrives than alienware now so its a no chance. m18x holds 3 HDD + optical + msata where as clevo probably just 3 HDD + optical.

x79 isnt even a 7 series chipset so it probably wont support trim in raid 0, not gonna get it. wait till IVY-E or haswell E, then we'll talk.

 

the above, and I already got x7200. I think for this year I'll build a desktop instead.

 

Lawl, I guess. Maybe there is a mSTATA on the revision? Also, the mSATA on the m18x is only a SATA II port, which is... kind of dumb. :/

I suppose, but IB-E will be the same socket as SB-E, so you could always just upgrade.

 

so you want the msata port to be sata III? then that means only one 2.5 inch sata port will be sata III rest will be sata II. so that means you will not be able to raid two of the same drive together, then whats the point in that?

all i am trying to say is, when R3 comes out it'll all be sata III ports, including the msata and thats counting on haswell mobile lynx chipset.

fyi, sandy mobile xm and ivy mobile xm uses same socket, same pins, but ivy mobile can only be used on 7 series chipset. so there you have it, ivyE and sandyE maybe same socket, but might not work and thats based on chipset/intel

 

I do believe that IB-E will be backwards compatible with the X79 chipset, but I can never know until it comes out and there is no guarantee that it will; I can see how that could be bothersome. The only article I've read that says it is compatible is Intel Core i7-3970X in Q4 2012; Ivy Bridge-E in Q3 2013 Yet that article may be out of date..
It is also troublesome that Intel has Haswell scheduled to be released before IB-E. If IB-E doesn't come with +6 cores I really doubt there will be any performance gain from using the enthusiast platform. Yet the IB-E die has up to 12 cores which gives at least some hope. Maybe the Xeons will have 12 and the desktop versions will have 10? One can hope. XD Nevertheless, if IB-E is not backwards compatible with SB-E I will be very sad, indeed.

And yes, I understand the limitations of the mSATA port on the m18x. As an aside I looked up the m18x r2 specifications and it seems like it has 2 storage bays and the mSATA while the P270WM has 3 storage bays. Correct me if I'm wrong though, I'm not entirely sure. X_X

 

hope this make it clear

 

Yup, totally agree with Unityole,

And yes, the M18xR2 has 3 storage bays + msata + optical drive = 5 hard drives. Personally, I don't care if msata is 3GB/s or 6. With a ssd in there you won't be disappointed. My personal dream setup would be: 2x512GB Vertex4 in RAID 0 + 1TB (octane) SSD for storing all my VM's and testing environment + 256GB mSATA Crucial M4 (alternative OS - Ubuntu 12.04 Server + VM's) + 1TB HDD in optical for backups and storage.

 

LOL holy cow thats one hell of a machine. 1tb octane is so expensive but its the only drive thats 1TB at 2.5 inch factor and theres no review on it, wish i could afford it though would definitely use them in raid for external storage lol.

regards to the multi threaded cpu performance drop as # of threads gets higher, can't recall where i find it but iirc 8 threaded cpu, each thread gets about 70-75% of performance compared to a single core CPU and that got me thinking more cores doesnt mean its better. more heat, less OC, less performance per core.. ouch!

 

Okay, I see. Thank you for clearing that up ^^

Why does the performance go down for each core, other than being indirectly caused by higher TDP,and thus lower clock speeds? I understand that the efficiency of more cores is not linear, but theoretically it should become more efficient, provided the software is available to use these extra cores.

If it is almost completely because of heat as a limiting factor, wouldn't it be logical to say that(assuming Intel doesn't use the cheap TIM again. -.-) IB-E would be capable of more cores while maintaining a similar TDP and thus be more efficient? Nevertheless, IB-E most likely will not scale/overclock as well as a the current SB-E processors since overclocking creates an exponential growth in thermal output; thus, more cores -> a larger exponent.

 

ok, found it here AnandTech - Making Sense of the Intel Haswell Transactional Synchronization eXtensions donno why performance drop as threads count increases but it does make sense if you think of SLI/Crossfire or Raiding SSD together you dont get 2x the performance, probably 25-50% at best. maybe its because the transistor count or maybe its because of heat, or maybe because of timing has to work with all cores, this will need an answer from intel themselves.

and regards to SB-E and IvyE, imho ivybridge is definitely better, cooler and faster and thats if you heatsink directly ontop of the CPU. ivybridge desktop CPU are crap because intel used thermal paste instead unlike sandy bridge, and this might carried over to IvyE too, never know.

Edit: so i just end up reading the article again and turned out multi threaded cpu needs to get information delivered correctly and something to do with lock anand talked about.

so good news is, if haswell cpu remove this "lock" maybe more threaded CPU wont lose that much of a performance, so look forward to haswell-E!

 

Actually, that article is why software runs slow. How would I know? Because I'm involved with some of those changes right now - writing finer tuned granularity for locking around shared resources within multi-threaded software applications.

However, there is a root cause for a slight slow down across these technologies - it is what I call overhead or bookkeeping. W/ CPUs cores, if a processes thread of execution alters memory, and that thread is moved to another core, all the register data must go along with that thread of execution. Keeping track of and moving register data along with a thread of execution is why you won't get exactly 2x the performance. Same would go for SLI/XFire - the GPU is asked to work on some calculation for an image, the data req'd for that may have to shift from memory on card X to card Y. Also the state of any current calculations would have to be tracked / moved as well. In regards to RAID, the performance will be X times in terms of pure IO. However, the book keeping of which disk the striped data lives requires a calculation. It so happens that since the SSD is so fast, these calculations take up more time than if the data just existed on one drive in the first place.

 

agreed. even with these lock removed, its based on software and software usually takes forever to catch up to hardware and probably reasons why i am using so many adobe programs as most are multi threaded.