AMD RX 9060 XT Specs & Price, Threadripper 9980X, 9970X, 9960X, & R9700 GPUs
jimmy_thang
May 21, 2025
We go over AMD’s Computex 2025 announcements which include the company’s new RX 9060 XT GPUs, Threadripper CPUs, AI Pro workstation GPU, and more
The Highlights
- AMD’s RX 9060 XT will have 8 and 16GB models
- AMD announced new Threadripper CPUs that include the 9980X, 9970X, 9960X along with PRO 9000 WX-series CPUs
- AMD also revealed the 9995WX, its new AI Pro workstation GPU, which will come with 128 AI accelerators, 32GB of GDDR6 memory, up to 1531 TOPS, and a 300W TDP
Table of Contents
Grab a
GN Tear-Down Toolkit to support our AD-FREE reviews and IN-DEPTH testing while also getting a high-quality,
highly portable 10-piece toolkit that was custom designed for use with video cards for repasting and water block installation. Includes a portable roll bag, hook hangers for pegboards, a storage compartment, and instructional GPU disassembly cards.
Intro
AMD just announced its RX 9060 XT GPUs (coming in two memory configurations). We already knew about these but the company just formally announced them. AMD also revealed its “Radeon AI Pro R9700” workstation GPU, and the company’s latest Threadripper 9000 series and Threadripper PRO 9000 WX-series of CPUs. Unlike NVIDIA, AMD actually wants people to know about its products rather than about anti-consumer, anti-free-press actions, so AMD not only announced the products and gave information on them, but will also be sending out review samples well in advance and without conditions - which normally isn’t worth mentioning, but is worth pointing out because of the recent NVIDIA issues.
Editor's note: This was originally published on May 20, 2025 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.
Credits
Host
Steve Burke
Camera, Video Editing
Mike Gaglione
Vitalii Makhnovets
Writing
Tannen Williams
Web Editing
Jimmy Thang
The 9060 XT 16GB will be $350 with the 8GB model at $300. They will release on June 5th. The GPU die is the same for both models, but new from the 9070 (read our review) and 9070 XT (read our review) (which also shared a GPU die). The new die is Navi 44 for the 9060 XTs and sized at 199mm^2, down from 357mm^2 on the 9070-class cards.
The company didn’t provide as many first-party testing results as they typically have in the past. That’s not necessarily a bad thing, because all of those results have to be taken with a grain of salt anyway, but we’ll mainly just be sticking to the specs today. We plan to review these cards once they launch. Our understanding is that, unlike the 5060, AMD plans to continue sampling GPUs as usual.
Just a heads-up: The information in this article is from a pre-briefing, so this is based on conversations with AMD and not the live presentation itself.
AMD Radeon RX 9060 XT
The 9060 XT will come in 16GB and 8GB versions.
As for the features that will be shared between the two: Each will have 32 compute units, 32 hardware RT accelerators, 64 of what AMD calls its AI accelerators, and a 3.13 GHz boost clock. Both models will run PCIe gen 5.0 x16 slots, DisplayPort 2.1a, and HDMI 2.1b. AMD also lists a range of 150W-182W for board power, which explains the single PCIe 8-pin connector pictured in the rendering of the GPU. In speaking with AMD, the lower end of the range is for 8GB models.
For reference, AMD’s 9070 XT has 64 compute units, 64 RT accelerators, and 128 AI accelerators, or double the amount of the 9060 XT’s CUs and accelerators. The 16GB 9060 XT matches the memory capacity of both the 9070 and 9070 XT, but with a weaker core.
These 9060 XTs will be direct competitors to NVIDIA’s 5060 Ti cards, even mirroring the same VRAM configurations. Despite identical memory sizes, NVIDIA’s 50 series cards utilize a newer GDDR7 memory compared to AMD’s GDDR6. As for how much that matters, that depends on the architecture and how much it’s going to rely on the memory bandwidth and that extra speed. We’ll look at it in testing and see how it performs in the real world.
Additional differences include the 9060 XT’s use of PCIe 5.0 x16 as opposed to the 5060 Ti’s (read our review) PCIe 5.0 x8 interface. In benchmarking at x8 vs x16 on gen 5, it’s not going to matter. The place where it might matter is socketing it into an older board where cutting the lane count in half is going to become a restriction in some configurations.
AMD’s first-party benchmarks compared it against the 8GB RTX 5060 Ti, which we think is fair since it’s a price-parity comparison. We’ll do our own benchmarking pretty soon in our review.
AMD Ryzen Threadripper 9000 and PRO 9000 WX-Series CPUs
Grab a
GN15 Large Anti-Static Modmat to celebrate our 15th Anniversary and for a high-quality PC building work surface. The Modmat features useful PC building diagrams and is anti-static conductive. Purchases directly fund our work! (or consider a
direct donation or a
Patreon contribution!)
AMD also announced its newest Threadripper 9000 Zen 5 CPUs, including the 9980X, 9970X, and 9960X, codenamed “Shimada Peak.” These have been upgraded with increased memory support and enhanced AVX-512 for more demanding tasks.
We’ll start with the standard AMD Ryzen Threadripper 9000 CPUs. This series includes the 9980X, 9970X, and 9960X.
The 9980X is a 64C/128T CPU at 3.2 GHz base clock and with 256 MB of L3 cache. The 9970X has 32 cores, 64 threads, a 4.0 GHz base clock, and 128 MB of L3 cache. And finally, the 9960X will come with 24 cores, 48 threads, a 4.2 GHz base clock, and also 128 MB of L3 cache. All of these CPUs will also feature an up to 5.4 GHz max boost clock, PCIe 5.0, the same sTR5 socket, and a 350W TDP.
For AMD’s Ryzen Threadripper PRO 9000 WX-Series: The company announced six new CPUs, which include the 9945WX, 9955WX, 9965WX, 9975WX, 9985WX, and the flagship 9995WX. Starting with the 9945WX and working our way up, these chips will come with core counts of 12, 16, 24, 32, 64, and finally 96 cores for the 9995WX. This mirrors the existing and prior 7000 series CPU configurations just now on Zen 5.
Both the PRO and non-PRO Threadripper CPUs seem to resemble the same basic specs as the 7000 series of Threadripper processors. In these spec sheets, with a higher max boost frequency for the 9000 series CPUs, but a lot of the rest is familiar.
One notable difference between the PRO WX and the non-PRO series of Threadrippers is that the workstation series offers “AMD PRO technologies,” which AMD describes as, “a robust suite of enterprise-grade features including multilayered security, advanced remote manageability, and long-term platform stability.” Additionally, at least in the past, the PRO WX-series cards supported the WRX90 chipset in addition to the TRX50 chipset.
AMD hasn’t announced any prices at this time, but the press-brief lists availability for July 2025, so we should be expecting to see these soon.
AMD Radeon AI Pro R9700
Finally, AMD introduced its latest AI Pro workstation GPU. Intel also just announced its new Pro GPUs this past week and we already have a tear-down up of the B60.
For specs, this RDNA 4 card will come with 128 AI accelerators, 32GB of GDDR6 memory, up to 1531 TOPS claimed, and a 300W TDP.
Compared to one of its predecessor, the Radeon Pro W7700, the new R9700 increases TFLOPS (FP16) from 56.54 to 96, increases AI accelerators from 96 to 128, upgrades to PCIe Gen 5 from Gen 4, and doubles the memory size from 16 to 32GB. Unfortunately, AMD’s press-brief didn’t include any CU, stream processors, or memory bandwidth info for the R9700, so we’ll have to wait to see those exact specs.
Due to the R9700’s noticeable configuration improvements over its predecessor, the new GPU ends up being slightly more comparable to the Radeon Pro W7800 which has 140 AI accelerators, 32GB of GDDR6 memory, and 90.5 (FP16) TFLOPS.
In its press-brief, AMD included a slide to illustrate how 32GB of VRAM gives users more options in their ability to load larger AI models by highlighting four models that would exceed 16GB of VRAM, but can be used with 32GB of VRAM instead. Additionally, due to the GPU’s ability to load models with larger parameters or that are less quantized, the GPU may also see an uplift in the accuracy of the model’s responses.
To expand upon that point, AMD includes another chart labeled “Large AI Models Performance” where it compares an RTX 5080 (read our review) to its AI Pro R9700. Once again, this chart demonstrates how 32GB offers access to run larger models that 16GB just can’t handle. These results are expected. We think a more meaningful comparison might’ve been using the RTX 5090 that also has 32GB of VRAM. This would represent a more like-for-like scenario but we don’t do a lot of ML testing so we’ll leave that for someone else.
AMD also shows off the card’s “Multi-GPU PCIe 5 platform” that allows users to connect 4 AI PRO R9700s for some extremely demanding models that need up to 128GB of combined VRAM and theoretically 4x the computing power.
We didn’t receive a price for this card, but AMD lists an availability of July 2025.
Conclusion
Visit our
Patreon page to contribute a few dollars toward this website's operation (or consider a
direct donation or buying something from our
GN Store!) Additionally, when you purchase through links to retailers on our site, we may earn a small affiliate commission.
That’ll wrap it up for AMD’s announcements.
Unfortunately, it’s a bit difficult to get an idea for performance based on the specs alone, and even harder to get an idea for the value for something without a price.
Ideally, we’d be able to get our hands on some of these once they’re publicly available, which should be soon according to AMD’s press-brief.
Connexion
