Taming the Beast: 4930mx

It depends on what you call overclocking. For any serious overclocking on the pre-BGA notebook CPUs only an Extreme CPU (XM and MX) can achieve any level of meaningful overclock. The rest of them are gimped just like the mickey mouse BGA K-joke filth, or even worse. The non-XM/MX CPUs are only good for a minor bump, like 400MHz or less, and they do not have fully unlocked power limits. Hardly worth writing home about.

 

"Overclocking" is defined generally as increasing the speed of the processor by any regards, and as such, MX isn't needed for overclocking. MQ/QM CPU's can enjoy a fair bit of overclocking headroom due to turbo bins, a 4900MQ can easily max out thermal/power limits of most machines (+6 turbo bins), and yes you can disable power limits on Haslel.
Overall, wouldn't call them gimped as a OC is still a OC. Reminder that there are BGA CPUs that will destroy MX/XM CPU's extremely easily.

 

I'd hope so, they are going on 4-6 generations newer so that is to be expected.

I will utilize the unlocked multi / potential binning / unlocked sbios / unlocked power limits / disabled power saving features for my overclocking methodology and for the purposes of this thread.

I already have the 4930mx chip and up til now confirmed that the 4980hq will not post in the ranger. It has a "locked" multi of up to x46 but is all moot as it will not post in the ranger even with Crystalwell microcode.

So, any thoughts of my question in my previous post?

 

A copper contact plate wouldn't net you huge gains, there are a small few to be had of which I assume you know of, and considering it is slightly bent, I'd go for the copper coated one

 

Copper coated?
What?

I have a copper contact plate, made of nothing but copper(including surrounding body), and the others are copper plate with aluminum body.

My heat exchanger can house 3 pipes in a single row, so with that in mind the copper might be the better choice just from a displacement of the body heat which isn't specifically hugely beneficial but I suspect may benefit from longer sustain loads.

Either way 3 heatpipes are happening here. Just curious what looks to be the better scenario.

If this project is a bust I'll still use it as a proof of concept to send to someone more capable than I for replication.

 

Haswell was a special case and "gimping" is due to firmware and hardware that the CPU is running with, not the CPU itself. That is what I was trying to explain to TheReceiver, ultimately performance will be determined by that.

4930MX vs 4700MQ overclocked


The 4700MQ overall beats the 4930MX and that really shouldn't happen. Ignore the PCMark04 score for the 4930MX, it's really a CBR11.5 score. BTW the 4980HQ overclocked absolutely crushes those scores but has an advantage with L4 cache, especially significant with some benches such as WinRaR scoring over 17,000.

@TheReciever, this is why for these chips the hardware and firmware is so important. Hopefully you can bring back some dignity for the 4930MX.

 

Yes I wanted the 4980hq for the L4 cache but sadly will not post. I have it running on my t440p now.

I have made the first heatsink, havent tested yet as I need to shave off some more heatsink to allow it to fit. The whole process, if I had done it correctly was rather easy.

Didnt bend much pipes, I just pulled pipes from other heatsinks and used those.

https://imgur.com/a/t2W0h4B


Give it some time as they are still uploading.

I am using the copper heat exchanger from the gpu side and also doubling it much like you see in the MSI laptops.

 

Holy snappin duck sh*t!
"Look at the size of that thing"


what...no water cooler

 

Haswell was the least impressive Core i7 mobile release in most ways. Only speculation, but I think Intel was not putting their best effort into it knowing that it was the last socketed mobile CPU with BGA to follow on its heels. The fact that the 4940MX CPU was an abject failure as an "Extreme CPU" (far worse than the 4930MX) supports that theory. The Haswell desktop CPUs were also rather pitiful. They ran extremely hot and the SoC idea was not executed gracefully. They were cramming too much crap on a chip. Sandy Bridge and Ivy Bridge XM, K and X SKUs were much better products, and those Extreme mobile CPUs distanced themselves appropriately from the ordinary consumer SKUs.

The other thing that has to be considered is the platforms the Haswell MX CPUs were being installed in left a lot to be desired. With Alienware, the firmware was totally crap, and with Clevos running @Prema BIOS their Achilles Heel was woefully inadequate thermal management. The whole Haswell era was an engineering abortion for all of the OEMs on the mobile front. It was not a good time for notebook overclocking enthusiasts. I'd say it was the worst that I can remember. What came before it was better for Alienware and Clevo, and what came after Haswell was amazing for Clevo (although things got a lot worse for Alienware).

 

Can you post the link with the 4700MQ results? And keep in mind that some benchmarks are very ram dependent. Also what OS you run can impact what results you get.

From the bot https://hwbot.org/hardware/processors?from=processor_2839#key=core_i7_4700mq

 

Unfortunately when deleting an account at hwbot it seems the results for that user get deleted too. Depending what you want to see then a google such as '4700mq hwbot wprime 0.0' might be enough or the webcache https://web.archive.org/web/20170708013959/https://hwbot.org/hardware/processor/core_i7_4700mq/

Doesn't seem to be much history for the 4980hq though


but enough, this has gone OT.

 

It's @Dufus. And he used desktop motherboard and custom water loop. Not apples vs. apples

As you can see... Some use exotic cooling. And some don't even mention it in the submission.

https://hwbot.org/submission/2517224_fugger_superpi___32m_core_i7_4930mx_6min_41sec_420ms

 

A desktop board was never used with the i7-4700MQ or the i7-4600M. Benches first were done with the laptop and stock cooling although was placed in front of an aircon for some results as daytime temps are over 30C most of the year. If in Norway could probably be done even better outside on a cold day. Later an Asrock IMB-180 mini-ITX board was used with temporary water cooling for experimentation. The 6" by 6" board is supplied by a single 12V-19V supply, has 2 SO-DIMM slots, a mobile PCH (QM87), LVDS for a display panel and a plethora of serial ports. The only resemblance to a desktop board would be the single 16x PCIe slot.

You can still Google some of the laptop only benchmark images of which results were posted in the vs image earlier.
https://d1ebmxcfh8bf9c.cloudfront.net/u23581/image_id_1417856.jpg
https://d1ebmxcfh8bf9c.cloudfront.net/u23581/image_id_1395840.png

Now the i7-4980HQ was used with a desktop board (Z97) and watercooled with warm water. With the 47W TDP processor reaching 200W with certain tests it was necessary.

No, and it's not supposed to be. It's to demonstrate what I have said from the start about the importance of hardware and firmware support. How will TheReceiver compare his own custom cooling?

Please remove your shoes before reading further.
The picture is not of his setup but his desktop wallpaper on his laptop. You may now kick yourself.

You can read about Fugger's laptop here
http://www.xtremesystems.org/forums...g-CyberPower-Fangbook-4700MQ-Corsair-Vengence

Do you remember the dual core mobile i7-4600M running air cooling beat all dual core desktop CPU's using XTU and took global first place for dual cores, even against those running LN2?
https://hwbot.org/news/12173_0_0_be..._with_air_cooled_mobile_haswell_in_2x_cpu_xtu

Not an apples to apples either but does demonstrate the hardware/firmware importance mentioned several times now.

Hope I haven't upset the applecart :/

 

Guys you can post what ever you want as the mods dont seem to care if this is Off Topic.

Ill take my results elsewhere in the forums.

4980hq is moot as it doesnt post in the Ranger. Im not sure why its being referenced.

4700mq is moot, as EC is not unlocked and does not support unlocked power limits in this platform. I have yet to see proof of this in the Ranger.

This thread is for overclocking the 4930mx IN MY RANGER, not overclocking in general, or even seeking advice about settings etc etc because others have already shared.

All of the methods used in referencial material required platforms that allowed it, or microcode exploits. None of which is desired or even intended for use.