Beelink GTi14 Ultra 185H Teardown 40 of 55 screws to reach the CPU

Подписаться 176

Просмотров 2,3 тыс.

50% 1

Synthetic tests data and graphs:
docs.google.co...
Beelink sent me a GTi14 Ultra and I have been working with it for the past few weeks. The performance is not as good as the previous SER8 but close enough I doubt the average person would notice outside of games. The main thing holding the mini PC back is the regular alarms from HWInfo64 telling me the CPU was thermal throttling even though the average temperatures were relatively good for a mini pc.
I was very surprised by the incredible number of screws inside this mini pc. I counted 55 screws in total without opening the power supply. 16 screws had to be removed to access the RAM and SSD, 40 screws were removed in the process of repasting the CPU. The thermal interface is an unusual mixture of liquid metal and thermal paste. Paper towel and 91% isopropyl work well to clean the cpu. It's just a time consuming process.
Aside from the incredible number of threaded hardware inside the GTi14 Ultra, Beelink has been making good effort to include new features for a mini pc like built in microphone, speaker, fingerprint sensor, power supply, vapor chamber, and an impressively large fan. If beelink could improve this design, it could really benefit from an easier method of accessing the RAM and SSD inside.
This is a mini pc that seems to be carrying a lot of on its shoulders with features instead of raw performance. Most people will be better served by the SER8 which is smaller, simpler, and has marginally better performance/cost. Unless you're specifically buying the GTi14 Ultra with its advertised GPU dock, this is a difficult product to recommend.

Опубликовано:

30 сен 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 19

@davidpower3102 18 дней назад

I wanted to share my experience with BeeLink support here. I’m based in Germany and I decided to return my GTi14 Ultra and Dock because of the thermal issues I mentioned here earlier. For the price and the size of the impressive cooler I was expecting more. Others however have reported much better thermal performance so I think I just got unlucky. So, back to the topic of Support. BeeLink support was very helpful. They got me an address to a local shipment company so my postal costs were cheap, and then when they confirmed the package was delivered they sent me my refund and also sent me a follow up link confirming the refund was processed. They were easy going, polite and quick. I’d definitely buy from them again considering how easy the return process is and how friendly support is.

@Power3DDesign 15 дней назад

I am really glad Beelink support was friendly with you and your return. I've heard some poor experiences from other users but Beelink was pretty friendly to me too when I requested a help with my GTR7 Pro last year. I paid shipping as well but was very happy they got me an orange replacement for the GTR7 Pro. I hope it is just a case that most people that had a decent experience with Beelink just don't talk about it much. I was expecting more cooling as well. I got some conductonaut liquid metal and if this doesn't fix the temperatures, I'm out of practical ideas.

@Srootus Месяц назад

Criminal how your channel has less than 60 subs, keep up the good work

@Power3DDesign Месяц назад

@@Srootus thank you!

@gaominliao320 Месяц назад

I bought the ultra 7 155H version of this machine. When the BIOS is set to 65W TDP, this machine only scored about 12,000 in multiple cores tests under the Cinebench R23 benchmark , which seriously does not meet the performance level of this CPU. I replaced it with two after sale, both of which were in the same situation

@Power3DDesign Месяц назад

@gaominliao320 that is very good to learn. My 185H was scoring 14,305 and I expect it to be a little better than the 155H but I agree these scores are lower than I would expect. According to cpu-monkey these new intel processors should be scoring between 16,000-17,000. Notebookcheck estimates a bit higher than that. It seems the thermal throttling of the CPU is happening with both the GTi Ultra 155H and 185H which is concerning. www.cpu-monkey.com/en/benchmark-intel_core_ultra_9_185h-cinebench_r23_multi_core www.notebookcheck.net/Intel-Core-Ultra-9-185H-Processor-Benchmarks-and-Specs.783353.0.html

@davidpower3102 Месяц назад

I wanted to jump in here too and share my observations. I’m seeing very poor multithread CPU performance on the 185H. I’m not sure what is going. I also got the AtomMan G7 TI which has the very same CPU and it is performing a lot better. I’m a bit disappointed to be honest. I expected the 185H to have better thermal performance than the AtomMan as it has such a big cooler. Have either of you found a solution?

@Power3DDesign Месяц назад

@davidpower3102 glad to know we are not alone and it seems like the 185H is just a super hot CPU even though the cooler is a massive chunk of copper. I reassembled the machine after lapping the heatsink but was not able to improve performance. I have some new liquid metal from thermal grizly in route later this week so hopefully using that instead of thermal paste makes a difference.

@davidpower3102 Месяц назад

@@Power3DDesign I don't get it either mate. The cooler in this thing is impressive, it should be to out perform the AtomMan X7 Ti but it isn't. Running the CPU Benchmark in 3DMark while running Core Temps shows that 5 cores thermal throttle a lot while running the tests. While only 2 cores thermal throttle on the AtomMan, and that's during the 2 and 1 thread tests. I am sad to say I will be returning the beautiful Beelink and keeping the odd looking AtomMan. This also means I need to return the GPU dock I have coming in the post... Great video mate, by the way is your sir name also Power?

@davidpower3102 Месяц назад

@@Power3DDesign I had some luck. I did a full clean “Rest this PC”. Set the power to 65watts and pushed the fan to an “always on full” mode - it still isn’t very loud (I had tested these BIOS settings before doing a full reinstall) I got noticeably better CPU scores in 3DMark. I also observed only 2 cores thermal throttling which isn’t so bad. The scores are still too low for this chip but it’s a marked improvement. My guess is there is some custom code or low level configuration which is interfering with the device. The fresh install seems to help a good bit. -- Update I posted an Update of my testing on the Reddit forum where people are discussing this video.

@JoelFabiani Месяц назад

Amazing. Thank you for the detained teardown. I hope 32gb is enough for my needs.

@618GOLDENRATIO 19 дней назад

Old school thermal paste, not a good idea, these run hot and throttle with the liquid metal, they used HIGH dollar liquid metal for a reason

@davidpower3102 18 дней назад

@@618GOLDENRATIO I did some reading on this when deciding to repast my Minisforum MS-O1. I didn’t find test results that conclusively showed that liquid metal was really that much better. Noctua’s NT-H2 (and a few others) still seems to perform well when compared to liquid metal. Full disclosure - I’ve never used Liquid Metal myself so I’m only reporting what I’ve read or heard on reviews. I was always curious to try it myself. Maybe it performs better is specific use cases.

@Power3DDesign Месяц назад

Sorry for the fan noise in the background! I didn't realize it was pretty bad. AC was out in my lab and I did not want to die.

@JoelFabiani Месяц назад

Oh my. Honestly I'd never put this together.