Hacker News Reader

Comment by psnosignaluk | Hacker News Reader

$499 for 12C/24T on the 3900X? $399 for 8C/16T on the 3800X? Those prices make this a very tempting chipset on top of the promised performance being touted by AMD. Looking forward to seeing the reviews and news on X570 boards. Hopefully the manufacturers will be plugging in decent features to match their threats of massive price hikes on AM4 boards with the new chipset.

View on HN

For a single GPU machine, I’m not sure what the use case is for X570 over a much cheaper B450 board. Most games currently aren’t GPU-bus bandwidth limited (or rather the GPU itself is the bottleneck) so I suspect PCIe 4.0 won’t impact benchmarks much.

Since 2 lanes of PCIe4 are as fast as 4 lanes of PCIe3 I think it can make sense for more IO if you don't want to go for Threadripper. On top of that the motherboards I have read about thus far have better VRM setups for the CPU so for overclocking that could make sense as well.

You do have to be careful that the B450 board you buy can handle the power requirements.

On all of these the GPU is directly connected to the CPU and there is no chipset in the way.

The benefits of X570 over B450 therefore have nothing to do with GPU performance but instead would be either overclocking capability or, more significantly, I/O to everything else.

B450 only provides 6x PCI-E 2.0 lanes and 2 USB 3.0 gen 2. That's not a lot of expansion capability, especially with nvme drives. Want 10gbe? Or a second nvme drive? Good luck.

X570 gets to leverage double the bandwidth to the CPU in addition to being more capable internally. So you'll see more boards with more M.2 nvme slots as a result, for example. And thunderbolt 3 support. Check out some of the x570 boards shown off - the amount of connectivity they have is awesome. That's why you'd get x570 over b450.

It still seems to me that B450 are perfectly adequate for a Desktop workstation/gaming PC.

Most people do not need a second nvme drive or 10GbE.

10 GbE or M.2 NVMe performance is already significantly degraded by being on a PCH in the first place. More hops, higher latency, much lower IOPS. Don't do it if you can avoid it.

The thing is that most things aren't (currently) bottlenecked by PCIe 3.0. A 2080 Ti shows about 3% performance degradation by running in 3.0x8 mode. 4 lanes of PCIe 3.0 is 4 GB/s (32 Gb/s) which is plenty for 10 Gb/s networking... or even 40 Gb/s networking like Infiniband QDR (which runs at 32 Gb/s real speed after encoding overhead). So you can reasonably run graphics, 10 GbE, and one NVMe device off your 3.0x16 PEG lanes.

And AMD also provides an extra 3.0x4 for NVMe devices, so you can run graphics, 10 GbE, and NVMe RAID without touching the PCH at all.

The real use-case that I see is SuperCarrier-style motherboards that have PEX/PLX switches and shitloads of x16 slots multiplexed into a few fast physical lanes, like a 7-slot board or something. Or NVMe RAID/JBOD cards that put 4 NVMe drives onto a single slot. But right now there are no PEX/PLX switch chips that run at PCIe 4.0 speeds anyway, so you can't do that.

> So you can reasonably run graphics, 10 GbE, and one NVMe device off your 3.0x16 PEG lanes

Sure but you won't find any board with a setup like that. You can also reasonably split the x4 nvme lanes into 2x x2 but again you won't find a such a setup.

You'll find no shortage of boards with everything wired up to the PCH, though, and it's "good enough" even if it isn't ideal. The extra bandwidth will certainly not be unwanted. Especially when you're also sharing that bandwidth with USB and sata connections.

> The real use-case that I see is SuperCarrier-style motherboards that have PEX/PLX switches and shitloads of x16 slots multiplexed into a few fast physical lanes, like a 7-slot board or something.

I think those use cases would instead just use threadripper or epyc. Epyc in particular with its borderline stupid 128 lanes off of the CPU.

I'd agree you probably won't see much of a performance gain once things are loaded when comparing B450 to the X570 boards, but using a PCIe 4.0 SSD will very likely improve boot and load times.

(I'm fairly certain for most gaming workloads, the bandwidth increase will only come into play when getting closer to 4k 144Hz, which is unlikely to be pushed out by first gen PCIe 4.0 GPUs.)

Rumors claimed there would be a 3700 that was 8c/16 64 watts and only slightly slower clocked than the 3700x. It was also supposedly $200 or so (the 3700x is $329).

Here's hoping it comes out, looks like a great CPU for a relatively cheap desktop.

If the 3700x is $329 I would bet on the 3700 being around $250, at least at launch. ~$80 cheaper is how they currently have it priced, 200$ for a slightly lower clockspeed seems too cheap, and it'd cut into their 3600(X) range.

But if you look at the launch price 2700 was $300 and 2700X was $330. I think the difference is so small that it drives people who don't want to manually overclock buy the X version to skip the hassle.

... If you can find any. I still can't find anyone to sell me a handful of AMD's embedded SoCs. I don't want to buy them on a board. I want to buy the chips themselves. I'm not sure what's up with that, if it's a supply issue or just a hard "no" to anybody that isn't an OEM.

This article isn't about embedded SoCs? It's about desktop CPUs that are readily available at retail. Embedded SoCs are a pain to buy from most vendors except the Chinese ones you can find on Alibaba.

My experience with AMD's supply problem is across more than just their embedded product line, but the supply problem is worst for me in their embedded product line. I'm sorry, I appear to have had some sort of mind frame shift recently, which is causing me to add layers of abstraction where inappropriate, resulting in communication errors.

Reply on news.ycombinator.com