BGA Venting Thread ;)

I think Dell can get away with it because the RAM is dual-rank, but you're right, this needs more investigation.

 

Those specs are always guaranteed and tend to get updated over time.

 

so would the 128GB work in 870TM is the question. can only ask @Prema

 

answer is no it won't.

also no handbrake?

 

no handbrake, have to wait until after 8 cores. most of the stuff i bought from 1-2 months ago from china are now arriving so obs will have to wait.

 

that MSI can support 128gb DDR4 Ram
https://twitter.com/EurocomTech/status/1030469437705596929

 

We need links, bro.
https://twitter.com/EurocomTech/status/1030469437705596929

 

looks amazing but 8086k? what about the 8 core support i'd assume it'll just work. so if 128gb works with 8086k it should work with 9900k. i wanna see the inside and the heatsink.. @Eurocom Support share some pictures pls

 

That's an MSI workstation.

 

Just go to X4C page and see 128Gb RAM option. As X4C, X7C, X9C is the same models, i thinks all of threm support 128Gb. Just the site is not updated yet.

 

I see no reason why it would not work.

 

thats the 17 inch eurocom claimed it was gonna come with 10nm 8 core icelake tornado F7 earlier this year or late last year. the vents looks nice something that only MSI would do, clevo doesn't seem to do vents on the end of left/right. if F7 only has 1 GPU then cooling must be pretty decent even for GPU as well but we'll see.

 

Pro stuff is insane! You can spec it close to $10K if you max the RAM and GPU, before adding nearly $14K worth of SSDs!

 

Yeah things like Xeons and 128gb of ram followed by 16tb of SSD can bump the price.

 

no need for xeon tho. no OC, expensive, all for the sake of ECC no thanks

 

I'm surprised AMD doesn't heavily capitalise on this fact. All their chips can support ECC

 

rofl im hoping they could talk companies into using their TR 32 cores in a laptop. because if it is only 8 cores, intels got those coming up and it has their performance beat.

 

I'd like to see the laptop that can handle 250 W power dissipation from the CPU alone...

 

just need to make it big and all good. 20 inch display laptop pls

 

How about a transformer laptop. I can see myself with a yellow/black laptop named Bumblebee.

 

See the p570wm, it was pretty close. Now make it bigger to account for the new GPUs so about 7.5kg. Shame no one will buy it.

 

If timing / price is right, where would I sign up for one?

 

count me in too.

thats a big no no, look at acer 21x sold more than 300+ units while charging a whopping 10k. surely people are out there, that doesn't even look at these forums. they don't need to go 10k, even 6-7k would be just fine for 500-1000 units.

 

Anyway back to the topic, before I sold my BGAlien 17, Dell changed the heatsink. And of course, the core differential isn’t fixed, isn’t that fantastic? :/

 

Some posts were removed.
Please keep it civil.

 

Even on here ownership of the P570WM was rare, in America in particular the purchase rate was low (Asia and Europe were OK). I was one of the few who had it. Clevo are understandably shy with it, even if it was down to timing.

It was pretty awesome with a hex core CPU, quad channel ram, dual GPUs and high refresh rate display options.

 

clevo has like 0 advertisement on it. im sure if they did as much as acer did it'll spread much faster. well all in the past, after this 8 core there will be no more laptop purchase for me for at least 3-4 years until something interesting come out. either 10-12c mainstream laptop or significant IPC/OC improvement.

 

With the BGA based throttling mess, timing would have to be spot on, but I imagine the response with a moderate footprint that can cool 165-200W CPU and could cool heavy duty 1080 from the LAN party and professional crowd may surprise them.

 

Thank you Mr Fox.

Already owned the BGAlien turdbooks. It was already so difficult to get 1275 points on an i7-8750H (and some know about the stupid tripod heatsink design on Alienwares).

With that being said, I don't know what to say about turdbooks, when there are bad business practices that laptop manufacturers don't want you to know.

Really hope when I re enter the gaming laptop section, LGA gaming laptops won't die . No one deserves lesser than what they paid for, even when people buy turdbooks.

 

Coffee LGA running i7-8750H BGA clocks (3.9GHz).

 

Awesome beans

 

they could have gone the route of 1 giant CPU, and a single GPU unit design. GPUs are so damn powerful these days dont really need SLI imho, majority people are settling with just 1 GPU tho. but then again thats me because im CPU/storage orientated.

 

Ditto! Just trying to appeal to the benching or gaming crowd.

 

Just saw this in my Youtube feed. Let's all run out and buy a BGA turdbook that is over-priced, over-sold by fanboys, and over-reminds us how stupid customers can be with their money. This is precisely why I joined this forum and why I am using my engineering background to recognize when a company is trying to sell me.

 

Physics does not care about marketing.

 

Neither does trust. Hopefully, the world we leave to our children will not be one where people only manipulate each other to gain the advantage..Gets old real quick.

 

So, I purchased a Dell Precision 7530 with a Xeon E-2176M inside. I am quite pleased with the performance and thermals:



All six cores pegged at 4.1 GHz, while running TS Bench, outputting a mere 60 W. A far cry from my old 4710MQ that ran at 3.5 GHz on all 4 cores, blasting out nearly 65 W...

Furthermore, I ran CineBench R15's multi-threaded test, twenty times. Attached is the resulting text file, and below is a chart:



1200 average score, easily 150 points ahead of most of the 8750H competition. Clocks during the CB15 run were maintained at 3.65 GHz constant. And temperatures never exceeded 85°C. Noice.

 

The power measurements like that are hard to compare.

 

Keep in mind of that 65W, ~10W was from the FIVR, which Coffee Lake does not have. That said, I'm not sure how your 4710MQ used 65W in TS Bench, when my 4720HQ used only 55W in Prime95 Small FFTs (AVX off) at 3.6GHz with a -50mV undervolt. Prime95 even with AVX off is quite a bit more power hungry than TS Bench.



On your first CB run, you're getting about the same score as an 8700K at 3.9GHz, so there still a little bit of throttling or performance limiting somewhere even before PL1 kicks in.

 

Is Dell using an offset? Is the sensor calibrated?

 

Perhaps it's the memory clock and timings? I've got DDR4-2666 CL19, whereas you've got (assuming that your numbers for the 8700K are from your own machine) DDR4-3000 CL16. Your memory has 3000/16 = 187.5 units of performance, whereas mine has 2666/19 = 140.3 units. Desktop DDR4-3200 CL14 is even faster at 3200/14 = 228.6 units.

Hence, your memory alone is 33.6% faster than mine; I'm fairly sure that that contributes to the increased performance of Cinebench R15, given that it's a ray-tracing benchmark and will likely need to access memory fairly often.

I also need to mention that for the moment, my memory is running single-channel, whereas yours (and most other bench setups) is dual channel.

 

Nope, I have an 8600K (no Hyper-Threading). I was going off of @Papusan's score, he gets about the same as you, at 3.9GHz. Like you said, it's probably the single channel RAM holding you back. CineBench doesn't really care about speed or timings.

 

CB does care about ram speed, to some extent. ram is basically treated as another layer of cache after L3, which in general it is and is slower than L3 cache. upping uncore frequency will up the CB score though less than upping core frequency ofcourse, upping ram will have similar but smaller results. @Ionising_Radiation did you undervolt your chip?

 

Ah, shucks. I've been running on low sleep ever since I started university. My bad.

According to this review, RAM channels don't impact Cinebench performance numbers, either...

 

Latency can though iirc.

 

From an older article... Ram influence: On the 6700k system dual channel 3200 to single channel slower 2133 only varied scores 1.5% or so. This with a faster memory kit in dual channel.

So from your link vs. the one I posted I would think minimum 0.8% and up to 1.5% loss with single rank memory. Fair to say you should be able to run your Xeon [email protected] BGA around +1% faster with the second 2666 ram stick. This means more like 1335cb vs. 1320cb on first round Cinebench R15. Maybe you could get higher score depending on how much you are able to undervolt your chips to lower Power consumption (To keep it well within PL2 limit). But it's quite cler your chips run into throttling territory.

Xeon E-2176M running 4.1GHz all cores witout throttling and dual rank memory should come closer to +1380cb.

But from what I have said several times... BGA will need minimum 1 bin higher clock speed vs. LGA chips to match or score right above. Anyhow how you put it... You should score above or minimum the results from 40x as showed below.

40x all 6 cores (no tweaked voltage for the BGA clocks).

 

cool idea, now we just gotta buy bga laptop to get those fans LOL

 

I thought it was a blade for slicing potato from the thumbnail. Maybe eagle wing or Falcon's wing design should give more airflow.

 

To be honest, the concept they're using in this part of the video is used in the GEnx jet engine as well: the engine nacelles are serrated, and have chevrons, precisely to make the mixing of the air emitted from the engine's bypass and the surrounding air, more laminar, thus reducing engine noise. The reduction was notable: here's a GE paper on the topic, from 2002.

We only saw the GEnx engines fly 9 years later, in 2011, on the Boeing 787. The concept, according to the paper, was tested as far back as 1996.

I feel that even a 10% improvement in fan airflow at a given RPM is a large leap over the previous; nowadays in the aviation industry, a 10% increase of airflow through an engine at equal RPMs (and hence almost equal noise and wear levels) is considered massive.

GEnx-2B jet engine on B747-8i:



The 'owl wing design' is probably more marketing smoke than real engineering, but overall, 10% CFM improvement is quite good.

I think we should keep pushing for improved notebook fan designs, from the motors, to the fan blades, to the location of fans from the chips themselves. Every drop counts. At some point we may even get semi-axial fans in notebook, which would massively improve airflow.

 

Would not have connected the two. I agree it's probably more marketing than anything but every little bit helps.