So of course, Gamers Nexus delidded it.
However, this puts more sense that the 11CU 2400G would run hotter with TIM with the stock Wraith cooler. Paul’s Hardware reported high temperatures when running in an open bench with the 2400G.
Budget option gets budget TIM
4 Likes
Wonder if der8auer will release a Delid Die Guard for the APUs. Might improve performace with the stock Wraith cooler.
It will probably cost as much as the apus…
1 Like
Not if he releases the STL for free use. I think this can be 3D printed.
2 Likes
A razor blade and solder are pretty cheap.
I’ve chosen the R7 1700 over the i7 7700K because I prefer my CPUs not to use toothpaste. I hope AMD won’t use this on the CPUs and stick with it on the APUs.
To be fair, Ryzen doesn’t run hot. So the toothpaste is most likely not limiting the maximum performance of those APUs.
With a stock cooler though, Paul’s Hardware with stock TIM was hitting 80+C. Wraith Cooler plus a delid and a delid die guard to allow direct contact between the die and the cooler will help for getting those temps down.
I bet any towercooler would fix that without having to delid.
1 Like
It appears der8auer’s delid tool for coffeelake works on AMD apus if you put the APU in upside down 
TIMESTAMPED
Takes a bit more pressure due to larger contact area
1 Like
I have a theory tech tubers dont know what to do with a bog stock APU that an every man could use and even game on. Tech tubers are so in a bubble that unless you put a GPU against it they can not make a youtube videos.
$100 dollars and its quad core and can game 720p and 1080p low settings befudles them.
Have a look at steam stats. Most people on steam have terrible PC’s. An APU like this will lift the steam gamer bar a lot for the everyman.
2 Likes
Well, it’s basically 1200 and 1400 with RX550… For less… Way less.
The only thing I can’t get over is the 4MB of L3 cache. If they improved that to at least 8MB I would be a little more interested in this APU.
The cache on this APU is fine at 4Mb. It’s still outperforming i5-7600’s etc.
The main reason the Static RAM for the cache was reduced by half was to reduce power consumption. and the size of the CCX footprint.
1 Like
I can understand it from an engineering and cost standpoint, but for a SMT processor, that’s not much to work with for each of the 8 threads. If SMT was off, then it makes more sense for the 2400G.
But the biggest gripe is the crippled PCI-E lanes. only x8 for discrete graphics which makes Crossfire and SLI impossible on X370, making a VERY BAD upgrade path for people.
I can see this processor for the mom and pap PC users for casual Netflix or HTPCs.
You fail to grasp what market segment this processor targets. This is not for your SLi/Crossfire enthusiasts or heavy gamers. This is a budget APU.
Bottom of the barrel from the raven ridge desktop lineup. It just happens to be really good for gaming too. It is built through compromises to a cost, not to an optimal engineering design.
The reduced cache causes almost no losses for this Chip in most use cases be it with or without SMT. Its also what makes this chip so cheap. Making large on-die SRAM for cache is expensive and complex and is the one aspect that most affects production quality. These CPU’s are minimally binned and built to allow for a very high yield rate at low production cost.
Messed up cache was the #1 cause behind the early Ryzen releases that lead to the segfault bug.
In fact if they where built with 8Mb of L3 and more PCI-e they would be an entirely new CCX design and cost far more to produce as well as have lower production yields and a wider product lineup due to binning. There could be dual cores selling at the $99 price instead.
Nobody buying this as a $99 or $169 CPU should ever be missing the extra PCI-e lanes. It should also be noted that it features 4 lanes for NVMe and another 4 lanes for the chipset. With an 8x segment equivalent of PCI-e IO used for the iGPU on the DataFabric thus leaving an 8x PCI-e 3.0 lane. Plenty even for a GTX1080Ti! At which point you would be sorely CPU limited.
This is why the upgrade path argument is silly for this as a budget part. This is not like the 2600k which was a high end part and had arguably the longest upgrade path of any CPU in history.
This was never EVER meant to be used in X370 boards. It just can be.
A320, A4X0 and maybe B350, B4X0, but really OEM machines is what this APU is made for.
The IHS being attached with TIM is also not surprising in the least. The majority of APU’s have never been soldered. Soldering is costly extra step you want yo avoid when making an affordable budget part.
9 Likes
It’s so much fun see people lose their shit over this.
"Why oh why this budget cheapo Cpu is not soldered?!?!?!11!"
The LM fix for it gives only 25mhz speed, but what it really give is better temps, and who’s going to fix it?
i ques ~0,001% of customers and every other youtube creator.
If the difference in price is around 5$ on soldered or tim, i would still take the cheaper one in this case.
I’m genuinely interested to purchase that cpu for livingroom media purposes.
Certainly not planning to delid for no good reason.
^This. Intel hasn’t used solder on their desktop chips since Sandy Bridge. Where have those people been for the past 6 years?
It’s astonishing how people get worked up over something so insignificant. This is silly.
2 Likes
Living without real competition in the CPU market of course.
Again, delidding could be worth it for ultra small SFF PC’s, like the one I want to mount to the back of my TV. Dropping ~10°C could absolutely be worth it in a system with exactly one fan (the CPU cooler) in it and a PicoPSU.
It’s not required of course, but is nice to have the option.
1 Like
You might want to wait for the rumored GE version.