- Joined
- Nov 15, 2021
Well, the larger memory chips still cost more money, as does integrating the larger package. Not using DIMMS does have disadvantages.
GPUs & CPUs & Enthusiast hardware: Questions, Discussion and fanboy slap-fights - Nvidia & AMD & Intel - Separe but Equal. Intel rides in the back of the bus.
- Thread starter Smaug's Smokey Hole
- Start date
Well, the larger memory chips still cost more money, as does integrating the larger package. Not using DIMMS does have disadvantages.
- Joined
- Aug 28, 2019
AMD Radeon 890M iGPU outperforms desktop Nvidia GTX 1650 in multiple games
Is 1080p finally conquered by iGPUs?
GPD claims AMD Radeon 890M iGPU will outperform Intel’s next-gen Arc A140V in TimeSpy
Not a bad showing for Lunar Lake tbh.
AMD partners shift focus to RDNA4, no sign of RX 7400/7300 models for now
The dream of an RTX 3050 6 GB competitor may already be dead.
RTX 4070 infused with slower GDDR6 pictured — New variant features 20 Gbps VRAM with 5% less bandwidth
It's now safe to upgrade.
Intel Raptor Lake 0x129 CPU Microcode Performance Impact On Linux
Some people are finding performance hits, but it's probably minor on average, with some specific workloads possibly taking a significant hit.
As always, there's a lot of mythology swirling around Apple. They have some people believing in magic.
Is 1080p finally conquered by iGPUs?
GPD claims AMD Radeon 890M iGPU will outperform Intel’s next-gen Arc A140V in TimeSpy
Not a bad showing for Lunar Lake tbh.
AMD partners shift focus to RDNA4, no sign of RX 7400/7300 models for now
The dream of an RTX 3050 6 GB competitor may already be dead.
RTX 4070 infused with slower GDDR6 pictured — New variant features 20 Gbps VRAM with 5% less bandwidth
It's now safe to upgrade.
Intel Raptor Lake 0x129 CPU Microcode Performance Impact On Linux
Some people are finding performance hits, but it's probably minor on average, with some specific workloads possibly taking a significant hit.
Pretty sure they can include double the capacity without changing the bus width, which is why M1 is offered with 8 GB or 16 GB and so on. They are also not using 512-bit on the low end. M1 is 128-bit, M1 Pro is 256-bit, M1 Max is 512-bit, M1 Ultra is 1024-bit. Same story for M2 variants, while the M3 Pro regresses to 192-bit. We also have "non-binary" chips available so an odd capacity like 12 GB LPDDR5X/LPDDR6 could probably be done on a 128-bit bus (like how we might see 12 GB GDDR7 GPUs using 128-bit within the next 1-2 years).
As always, there's a lot of mythology swirling around Apple. They have some people believing in magic.
I wouldn't be surprised if Apple is using 2x 4 GB or 8 GB LPDDR chips that are the exact same size as each other to reach 8-16 GB on M1/M2/M3/M4, but I'll have to research that because I don't know. Changing the package size between two different RAM variants of the same chip seems like a non-starter.
- Joined
- Jul 18, 2019
It will likely have the same amount of ports it currently does. The issue with the current case is that it's the same one they've been using since 2010 and is actually comically oversized for the M-series SoCs. The cases on current Mac Minis are more than half empty space:
- Joined
- Mar 30, 2023
It will have four Type-C ports, a hdmi port, and possibly an ethernet port. Nothing else
The low end model will have two Type-C ports
- Joined
- Sep 15, 2016
They can squeeze the M4 with oodles of RAM into the new iPads.
So it can get a LOT smaller than that. Rumors say something the size of the current Apple TV which is plausible.
If a little silly. Once you hit Mac Mini size, size doesn’t really matter that much. I’d rather they stuck a proper fan/heatsink on there and enough ports.
- Joined
- Mar 30, 2023
It would be nice if the Mini was compatible with OPS ports, or small enough that you could fit it in an adapter for one.
- Joined
- Mar 30, 2023
Would anyone ever consider buying used cpus off Aliexpress?
When i rebuild my server I want to make it fast enough that it can run a Windows VM on top of it, and still be able to hold a lot of SAS drives
When i rebuild my server I want to make it fast enough that it can run a Windows VM on top of it, and still be able to hold a lot of SAS drives
Leaded Gasoline
kiwifarms.net
- Joined
- Dec 7, 2023
I have bought a bunch of old Xeons off Aliexpress and all of them work fine
this 2696v3 I'm running right now is likely the best price / performance CPU you can get and it's only $50
- Joined
- Nov 15, 2021
CPUs last forever (except 13th and 14th gen Core, lmao). Memory DIMMs are usually the first thing to fail (or the PSU). So if you're a server with an old CPU, and make sure the first-to-fail parts are new, you should be good to go.
- Joined
- Mar 30, 2023
What about something like this? it would let me plug in a SAS controller, two video cards (an Intel Arc for transcoding and a Nvidia Quadro for the Windows VM) and two m.2 slots.
then i would just need a PSU and could plop it into my existing case
Leaded Gasoline
kiwifarms.net
- Joined
- Dec 7, 2023
https://xeon-e5450.ru/socket-2011-3/machinist-x99-mr9s/
It looks like the stock BIOS does not have overclocking features but it's a great kit if you just want it to work
- Joined
- Mar 30, 2023
I realized that one of the full length PCIe slots is axtually x1 and not x16, so that's a no-go as i need two x16 slots for the gpu and at least a x8 slot for the SAS controller.
this one has three PCIe x16 slots so it will work, but I will have to get a pretty tiny ssd for the stubby m.2 port
- Joined
- Aug 28, 2019
"Actually, we found that Zen 4 is sometimes more efficient than Zen 5..."
Just Some Other Guy
kiwifarms.net
- Joined
- Mar 25, 2018
Gamers still whining that these chips weren't aimed at them?
They told AMD they were waiting for X3D chips. So they might as well design these for a different segment.
- Joined
- Jul 18, 2019
They have to wring as many vids out of this week as possible since 9900x and 9950x are launching on Thursday and will probably perform better, thus letting them do a completely different series of vids along the lines of, "WERE WE WRONG ABOUT ZEN 5?!?!"
Between this half-launch, R9, and X3D later this year/next year, that's a solid 3-4 weeks of Zen 5 coverage AMD has handed them to milk videos from.
- Joined
- Dec 19, 2022
I've done this thrice, it's fine. Each chip arrived clean and working fine. Be wary if you're buying pin grid array, I've seen pictures of people who received the processors in simple padded envelopes, obviously with every single pin either bent or broken off, but land grid array should always be fine even if they're not packed properly.
You can still buy PGA of course, just make sure the seller puts it either in the original packaging or in a purpose-made tray. Basically just read their reviews.
- Joined
- Aug 28, 2019
Sounds like AMD is softly crying:
AMD Ryzen 7 9700X & Ryzen 5 9600X “Zen 5” CPUs To Receive New “105W” TDP In AGESA 1.2.0.1a BIOS Update
AMD set to boost TDP for Ryzen 5 9600X and Ryzen 7 9700X to 105W with upcoming AGESA update
If that's true (source is an anonymous Twitter leaker), it would be unprecedented for a shipped product AFAIK. It wouldn't be coming soon enough to affect the initial wave of reviews, so it could be considered doubly stupid.
I'll lead out the 9950X/9900X reviews with Chips and Cheese since they got one out uncharacteristically quickly:
AMD’s Ryzen 9950X: Zen 5 on Desktop
Last edited:
- Joined
- Jul 18, 2019
Barely faster than 7950x *and* it requires game bar and core-parking for gaming otherwise it's slightly worse than the 7950x. Apparently the cross-CCD latencies on 9950x are atrocious:
Incredible.
- Joined
- Aug 28, 2019
That agrees with Chips and Cheese (not going to bother converting the webp). Very strange. I previously expected Strix Point to have latency weirdness from having two CCXs and a 4x Zen 5 + 8x Zen 5c design, but it also seemed bad despite being monolithic:
It could be a quirk of how Zen 5 works.
Last edited:
- Joined
- Nov 15, 2021
Just looking into GN's numbers a bit. I think he is right that the efficiency gains here are pretty underwhelming. One reason is there supposedly is only about a 6% transistor density improvement going from TSMC N5 to N4P. (This is what people are citing: https://www.techinsights.com/products/ace-2204-801). TSMC claims a 22% improvement in power efficiency.
But he is also directly comparing unlike processors. The higher you clock a CPU, the less efficient is, so it is not surprising at all that the 9950X, with its clock of 4.3-5.7 GHz, is less efficient than the lower-end CPUs with slower clock speeds. he doesn't mention that in his 7-Zip section. Notably, the 9700X (3.8-5.5 Ghz) is quite a bit more efficient than the 7700X (3.8-5.3 GHz), about 38% more efficient. That's not nothing.
But he is also directly comparing unlike processors. The higher you clock a CPU, the less efficient is, so it is not surprising at all that the 9950X, with its clock of 4.3-5.7 GHz, is less efficient than the lower-end CPUs with slower clock speeds. he doesn't mention that in his 7-Zip section. Notably, the 9700X (3.8-5.5 Ghz) is quite a bit more efficient than the 7700X (3.8-5.3 GHz), about 38% more efficient. That's not nothing.