Thursday, February 15th 2024
AMD Ryzen 9 7900X3D Drops to $409, to Clash with Core i7-14700K
AMD Ryzen 9 7900X3D is the often-ignored middle child of the 7000X3D series that's flanked by the reigning gaming CPU champion, the Ryzen 7 7800X3D; and the company's flagship Ryzen 9 7950X3D, which performs within 5% of the 7800X3D in gaming, but with the added 8 cores shoring up its productivity performance against the Core i9-14900K. Pricing of the 7900X3D dropped to $409 on Amazon, which is a huge departure from its $600 launch price. At this price, the 7900X3D is set up for a direct clash with the Intel Core i7-14700K, which is going for $400, with its iGPU-disabled sibling, the i7-14700KF listed at $392.
The Ryzen 9 7900X3D is is a 12-core/24-thread dual-CCD processor, with its 12 cores spread among two CCDs in a 6+6 configuration. The first of the two CCDs has the 96 MB L3 cache thanks to the 3D Vertical Cache (3D V-cache) technology, while the second is a regular CCD with just the 32 MB on-die L3 cache, but which can sustain higher clock speeds than the 3D V-cache CCD. The similar 16 core 7950X3D flagship can be had for $600, or about $50 higher than the i9-14900K, while the 7800X3D is going for $370.
Source:
VideoCardz
The Ryzen 9 7900X3D is is a 12-core/24-thread dual-CCD processor, with its 12 cores spread among two CCDs in a 6+6 configuration. The first of the two CCDs has the 96 MB L3 cache thanks to the 3D Vertical Cache (3D V-cache) technology, while the second is a regular CCD with just the 32 MB on-die L3 cache, but which can sustain higher clock speeds than the 3D V-cache CCD. The similar 16 core 7950X3D flagship can be had for $600, or about $50 higher than the i9-14900K, while the 7800X3D is going for $370.
153 Comments on AMD Ryzen 9 7900X3D Drops to $409, to Clash with Core i7-14700K
The only "fixes" M$ have added to the OS/game bar etc is better game detection so that the behaviour of locking the relevant CCDs is correct more often. When it was day 1 there was quite a few times when you would be "gaming" but the OS thought you were in "productivity" so would shift all processing to the non X3D CCD.
It was the one faux pas AMD did with X3D that Intel did better with P/E cores. AMD relied purely on a software based solution where Intel actually did hardware based scheduling in its silicon. AMD would either have to equip X3D chips with both X3D equipped CCDs or create a new I/O die revision with a new hardware scheduler in it to mitigate this completely.
So maybe you wanna get your facts straight before claiming BS.
The issue though is that no one who is arguing has the actual chip. Just because you have a 7600 does not mean that is as far as it goes. Let's try an experiment since you have a different CPU. Let's get a Game at 4K that is CPU bound like TWWH3. Then let's see at 4K where the FPS in a Custom Battle using the same settings.
That will put to bed the 0% difference that you claim. As an example the 7900X3D feeds the GPU 3-5 GB/s vs the 5800X3D. I will be buying a 7900 soon and will put that to the test too. Don't forget Pcie 5.0 as well.
This also shows why 4K game testing for CPUs is sub optimal. I assume the Intel chips have more stable framerates, because in situations where the system is CPU limited not GPU limited, the 7800X3D is faster.
3D cache benefits 16 core CPU with ~16% uplift, 12 core CPU with ~10% uplift and 8 core CPU with ~17%
People upgrade their GPU more frequently than their platform in my experience, so CPU isolation testing is always useful.
hardware/comments/12hth3b
^ meta review in full.
I also do not take anything that he talks about very seriously while he is talking about a day one review.
Thread assignment shenanigans make hybrid CPUs (yes intel too) very annoying to obtain consistent/accurate results.
Not all tests show the 7950X3D slower by default, so incorrect.
Having owned and thoroughly tested a 7900X3D, disabling the non vcache CCD, strictly for testing purposes, it is indeed faster than a 7800X3D. It’s a shame thread hoping across CCDs can hamper performance.
*edit for some context while testing CP2077 awhile ago
Disabling the non vcache CCD resulted in +15% avg FPS and +20% min FPS
docs.google.com/spreadsheets/d/1TAvRLbaLzooI63cT0rzkHvIy3un8GbCUT0oZ6a0qxy4/edit
More relevant is what happens in 720/1080p, as this is CPU-bound scenario.
In 720p, 7900X3D would land ~5% slower than 7800X3D, so somewhere near 7950X3D/14900K
When scheduling works, and there's no issues with latency, you're right, the 7950X3D is faster than the 7800X3D, the problem is it doesn't always work in reality, as testing shows.You can undervolt non X3D chips to get the same efficiency gains.
The X3D chips can undervolt further because they don't have to hit the same frequency targets. You can't expect a 5 GHz chip to undervolt to the same voltage as a 4.5 GHz chip, while maintaining performance.
Further details from 3DCenter just confirm that price correction was the right move, as it compensates for smaller v-cache uplift on 7900X3D.
At ~$400, now suddenly this CPU is back to life and very competitive.
The 7900X3D probably should’ve never been a thing without better thread assignment, which makes it look a lot worse than it actually is when it comes to gaming.
Reminds me of when AMD sold dual CCD 3+3 5600 and 4+4 5800 without telling the consumer or changing model numbers.
www.techpowerup.com/277053/dual-ccd-ryzen-5-5600x-and-ryzen-7-5800x-in-the-wild
At least the 5900X is faster in it's intended role - productivity, than the 5800X.
Not to mention with proper thread assignment there are games that primarily run with a main thread or dont fully utilize all cores/threads on the 7900X3D/7950X3D that are faster already.