Low Spec Gamer is a great channel.
And sometimes…wahahahaha
7 Likes
looks OK enough to make me feel the PS5 will be worth it on their 12nm process in a year and a half but I’m still much more interested in the mobile version of this product and since AMD is not likely to compete on power consumption I’m mostly just hoping this puts pressure on Intel
reviews so far suggest that the 2200G can easily outdo an i3 apu, and if that pricepoint of 100 bucks is true, the questions that remain are
- where to find cheap RAM,
- a mATX mainboard that offers enough SATA interfaces for at least six discs (alternatively a pciX card, since no gpu is necessary?)
- and if it is possible to run this cpu without a fan/just using a massive heatsink.
haven’t found reliable tdp-figures that go beyond quoting AMD presskit.
anandtech is reporting ~60w, which is borderline for just a massive heatsink
hm, OK!
either gotta buy another noctua horizontal heatsink then, since i absolutely won’t go bigger than mATX, and those discs must go somewhere.
also AMD and Intel have now had four years to even get close to Nvidia on GPU perf-per-watt which is kind of a long time 
exactly what kind of proprietary unduplicable magic did maxwell entail
then again it seems like Samsung are only due to catch Apple in the same metrics for ARM CPUs this year
I’m cool with Apple and Nvidia actually keeping moore’s law alive in their respective sectors, it’s just troubling that the two of them have had no competition for as long as Intel has been treading water
actually Apple’s iOS GPUs seem more in line with Nvidia thermals too
(discourse is breaking link)
gfxbench dot com/compare.jsp?benchmark=gfx40&did1=52778367&os1=iOS&api1=metal&hwtype1=GPU&hwname1=Apple+A11+GPU&D2=Intel%28R%29+Core%28TM%29+i7-7600U+CPU+with+HD+Graphics+620
I’m not going to stan for Intel but Nvidia’s R&D budget is more than AMD’s market cap. It’s a miracle they haven’t fallen further behind.
1 Like
https://selectbutton.net/user_avatar/selectbutton.net/notbov/240/46836_1.png
anyway, both Raven Ridge desktop APUs are 65w TDP parts. basically, every desktop Ryzen SKU has a 65w TDP except for the X SKUs, which have 95w TDPs. the numbers I’m seeing under load is 100w for the 2400G at stock clocks from the wall, which probably puts it in the 70-80w zone.
The main competitor to both Apple’s CPUs and Nvidia’s GPUs is Qualcomm, not Intel. Not that Qualcomm has really been doing better than good-enough either though
Both Ryzen and Vega were designed to be low power. And by that I mean, their power/performance ratio is supposed to be best in laptop/notebook and such related hardware. However, the process isn’t refined/flexible enough to scale up to meet the highest desktop performance. And AMD is still trying to get a full suite of products out. So, I think Ryzen and Vega will do well in the low range notebook and professional notebook sectors. Their hybrid APU with Intel could be a thing, who knows. But, I’m not sure that AMD will have the time to get out their own performance mobile product. So, gaming laptops will still probably end up with separate, discrete Nvidia graphics, for the near future. it seems like they could easily spinoff the chip from the Intel hybrid, as a mobile discrete chip to compete with Nvidia. But I’m not sure if AMD is interested in that. What with their heavy investments into APUs.
It seems pretty clear that both Ryzen and Vega are intended to carry forward with AMD’s console projects. Indeed, I think some variant of Ryzen 2 will end up on next gen consoles. But I don’t think Vega is going to live that long. I think that Xbox One X will be the only console application of Vega. By the time Ryzen 2 is ready for a console variant, I think AMD’s true next gen GPU architecture will also be ready/ready enough, to shut out Vega. Even if it means some sort of half baked in-between product.
12nm isn’t a true, complete die shrink, like previous shrinks. Only parts of the die are shrinking. So its really just a marketing thing, which actually means a 12nm/14nm hybrid. That alone, won’t really be enough to mean much for Vega. Its likely they will sneak in some process/silicon refinements, along with it. So we might actually have some solid overclocking Vegas in the refresh. But I don’t think we will see any big cuts to power usage. Who knows, maybe it will be enough to release a new higher SKU, like an Vega 86 or something.
Intel announced a couple of months ago that they are getting into discrete graphics cards, which could result in high end gaming cards (I would bet at least 2 years away), among other high end applications.
They snatched Raja Koduri from AMD, who had been the graphics lead for some time. Its unclear if Vega is what it is because of his leadership (so, kinda bad), or if he managed to make it worth anything at all, despite possible budget/resource constraints at AMD. I dunno. When it was first announced he was leaving, there seemed to be an air of relief in some computer communities on the web. Implying that he may have effed up with Vega. I don’t even pretend to have a personal take on that. But, Ryzen is doing well for AMD, financially. So hopefully that money can reinvigorate their graphics department.
Raja is at least decent at managing graphics projects. With Intel’s unlimited money, I have no doubt their eventual products will be pretty good. Intel’s graphics are already pretty good, for what they are. and they have done a good job of leveraging them for added features to do video decoding, encoding, capture, and streaming.
The upcoming “Navi” from AMD, is really only an evolution of Vega. Navi could actually be scrapped, since it would be pretty close to the actual new architecture design (Not GCN based). And if Ryzen continues to trickle in profits, I could see such speculation becoming a reality (especially with new leadership who may not feel attached to Navi). And I think it would benefit AMD to be aggressive and go ahead and get on with their new achictucture. Set it up for console, mobile, and desktop computing, all at once.
1 Like
that’s a good point! it’s just a bummer that there’s no product (reasonably priced or no) that can match that efficiency in certain market segments where it really matters (namely, laptops), and (if you’re as mindful of how good things should be as I am, which is not really recommended) this has been the case for a little while.
Well, the mainstream computing mode today is phones connected to server farms, leaving everything from tablets to desktops as an elites-only high-end productivity and entertainment niche. It’s basically what Steve Jobs said about PCs being like “trucks”, although he didn’t realize the analogy would also fit tablets.
It’s still a decent-sized market but it only makes sense to focus core R&D fixed costs on phones or servers and then make derivatives for the mid-sized devices, which is what we’re seeing. You can only hope that the technology developed for those other things will happen to scale well to the niche.
1 Like
this is admittedly true (and is a useful way of taking PC enthusiasts down a notch) but insofar as I still need a laptop it’s not really relevant imo? It’s the kind of thing I expect a Microsoft VP of international development to say – sure, all these new markets are doing yadda yadda yadda and it’s worth being mindful of that, but … that has nothing to do with me as an end-user. their standards are much different than mine, for obvious reasons.
also, broad statements like that ignore that there’s been a tiny bit of market share going back to Windows desktops (to my mild dismay) because of how powerful consumer GPUs have become and how notebook form factors haven’t really had the products they need for the past few years.
interesting notes about Ryzen 3 2200g and Ryzen 5 2400g:
Improved 14nm process and other refinements are an early preview of Ryzen+ (Not Ryzen 2) with better stock clocks and refined clock steps.
L3 cache is cut down. But won’t be an issue most of the time, due to clock increases Vs. similar Ryzen quad core products.
The regular Ryzen quad core cpus had two CCX controllers in them (parts which facilitate communication between cores and was an early topic of latency issues in the high core parts). These new APU consolidated CCX duties into a single, 4 core unit.
A. this eliminates potential latency problems, althought its probably not an issue in quad core parts anyway.
B. Apparently, unifying to one CCX design, saved so much power, that the GPU being present does not really increase power usage and also made enough room in the die. So the die isn’t larger. Its probable that using less cache helped here, as well. Less cache also probably help with price.
C. These APUs only allow for a PCI-E 8x lane for a discrete graphics card. That’s probably fine, for this level of CPU. and limiting lanes could also have helped with power usage.
I saw some very detailed tests a couple of years ago and a 980ti only suffered a couple of frames per second, on an 8x lane Vs. a 16x lane. So, a 1060, 1070, or 1080 non-ti, would also likely be fine. As well, RX 570, 580, or Vega. Dual and triple cards suffered noticeably. Especially triple. I’m unsure how some like a 1080ti or the newest Titan variant, would fare on 8x. But its unlikely someone would pair those with one of these lower Ryzen, anyway.
oh geez! that blows!
like I guess it’s probably tolerable for most of the use cases they’ll be put to but I got a 3570k in 2013 and didn’t get a GPU for it for three years and was very happy to benefit from the full power of that GPU when I did because it is a Titan X
(did not pay out of pocket)
1 Like
And another note:
The GTX 1030 from Nvidia (Which Ryzen 5 2400G is basically equivalent to in graphics performance), is offcially stated as 30w tdp.
Regular Ryzen 5 is 65w TDP and Ryzen 4 2400g is also 65w. So, you can get an idea for the margins they moved, to fit into the same envelope, and result in a mostly faster CPU + a GPU basically equivalent to a GTX 1030.
do you realize that by the time you get enough GPU for PCI-E 3.0 8x to make a difference you’re going to be CPU bottlenecked because you paired a 100-150 dollar CPU with an 800-1000 dollar GPU
1 Like
barely though tbh. $150 CPU and $800 GPU actually sounds about right to me as far as the CPU not being the bottleneck
Another way AMD kept the cost down on these Ryzen APUs: The IHS is not soldered to the die, like the regular Ryzen CPUs. They used a thermal past, like Intel does.