Asus unveils AI Cache Boost — claims up to 19% faster AI workloads on Ryzen 9000 series
The new feature arrives as part of a new BIOS update.

Asus has unveiled a new BIOS feature called AI Cache Boost, which is designed to enhance AI performance on systems equipped with AMD’s Ryzen 9000 series processors. The new feature is part of a recent BIOS firmware update for Asus AMD 800 Series motherboards. It focuses on improving the efficiency of data transfers between the CPU cores, cache, and memory, which is crucial for handling extensive AI workloads.
Upon enabling AI Cache Boost in the BIOS under the Extreme Tweaker tab, the system overclocks the Infinity Fabric clock (FCLK) to 2100MHz. This increase in FCLK enhances the bandwidth of data transfers within the CPU architecture, which Asus says facilitates more efficient processing of large datasets commonly associated with LLMs. As per Asus, enabling AI Cache Boost can boost performance by up to 12.75% when working with LLMs (Large Language Models).
This feature is tailored for AI enthusiasts, researchers, and professionals who utilize AMD Ryzen 9000 Series processors, specifically those with AMD 3D V-Cache Technology, like the recently launched Ryzen 9 9950X3D. Asus says it is particularly beneficial for users engaged in AI workloads that involve processing massive datasets, such as training and deploying large language models.
Asus has clarified that the new AI Cache Boost feature is specifically designed to enhance AI workloads, not gaming performance. While the feature can technically be enabled alongside other BIOS settings like GPU Boost, doing so may compromise overall system stability without delivering any noticeable benefit in gaming scenarios.
For users focused solely on maximizing AI capabilities, Asus offers additional tuning options beyond AI Cache Boost. By activating both Game Turbo and AI Cache Boost modes, disabling simultaneous multi-threading, and deactivating one CCD (Core Complex Die) on Ryzen processors, Asus says AI performance can be pushed even further. In fact, Asus own tests on a Ryzen 9 9950X3D configured with a single eight-core chiplet showed up to a 19.4% improvement in AI tasks. As with all company-provided benchmarks, take that with a grain of salt.
Asus conducted internal testing using a system equipped with ROG Crosshair X870E Hero motherboards, Nvidia RTX 5090 or 4090 GPUs, and Ryzen CPUs with AMD 3D V-Cache, showing performance uplifts in various AI benchmarks. In Geekbench AI tests, enabling AI Cache Boost resulted in improvements between 3.4% to 8%, with the Ryzen 7 9800X3D demonstrating the most significant gains. The uplift was similarly positive when tested with previous-gen RTX 4090 GPUs, showing a boost of up to 7.57%.
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 43932 | 46213 | +5.19% |
Half Precision Score | 63464 | 67009 | +5.59% |
Quantized Score | 32197 | 33347 | +3.57% |
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 43625 | 46074 | +5.61% |
Half Precision Score | 62772 | 66744 | +6.33% |
Quantized Score | 32169 | 33275 | +3.44% |
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 44099 | 46965 | +6.50% |
Half Precision Score | 63365 | 68393 | +7.93% |
Quantized Score | 32480 | 33774 | +3.98% |
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 37480 | 40181 | +6.19% |
Half Precision Score | 51786 | 54392 | +5.03% |
Quantized Score | 28097 | 29118 | +3.63% |
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 37866 | 39777 | +5.05% |
Half Precision Score | 51782 | 53642 | +3.59% |
Quantized Score | 28157 | 28798 | +2.28% |
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 38191 | 40788 | +6.80% |
Half Precision Score | 52531 | 56509 | +7.57% |
Quantized Score | 28247 | 29536 | +4.56% |
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
Single Precision Score | 38277 | 40441 | +5.65% |
Half Precision Score | 52465 | 55823 | +6.40% |
Quantized Score | 28373 | 29284 | +3.21% |
Asus testing using the UL Procyon AI Computer Vision Benchmark the Ryzen 7 9800X3D achieved double-digit improvements of up to 12.75%. As mentioned earlier, Asus says combining AI Cache Boost with Turbo Game Mode can yield even higher gains. For example, the company says the Ryzen 9 9950X3D saw up to a 19.35% performance increase in specific AI workloads when both features were enabled. This combination is particularly beneficial for AI users whose tasks don’t fully utilize the CPU's multithreading capabilities.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Row 0 - Cell 0 | Default | AI Cache Boost | Improvement |
AMD Ryzen 9 9950X3D | 1426 | 1485 | +4.14% |
AMD Ryzen 9 9900X3D | 1417 | 1473 | +4.00% |
AMD Ryzen 7 9800X3D | 1490 | 1680 | +12.75% |
AMD Ryzen 9 9950X | 1439 | 1475 | +2.50% |
Row 0 - Cell 0 | Default | AI Cache Boost | Turbo Game Mode | Improvement |
AMD Ryzen 9 9950X3D | 1426 | 1485 | 1702 | +19.35% |
AMD Ryzen 9 9900X3D | 1417 | 1473 | 1570 | +10.80% |
If it works as advertised, the AI Cache Boost features could be a straightforward way for Ryzen 9000 series users to extract more performance out of their systems when working with AI-related tasks.
Kunal Khullar is a contributing writer at Tom’s Hardware. He is a long time technology journalist and reviewer specializing in PC components and peripherals, and welcomes any and every question around building a PC.
-
bit_user So, can anyone confirm whether it was CPU or GPU inferencing? The article mentions a RTX 4090, but where I'd expect this to have a real impact is on CPU inferencing. Perhaps there are some CPU-based layers that need to be processed, so it's a hybrid situation?Reply
It did seem to me like Ryzen 9000 is being held back by its IF links, so this doesn't surprise me. It'd be nice if AMD would do a 9000 refresh with a new I/O die, to give us faster memory support + faster IF, this year. However, if that were in the cards, I think the rumor mill would already be buzzing about it. I guess we'll have to wait until Zen 6 for a proper fix to these problems. -
setx Setting FCLK to 2100+ is obvious tweak to boost memory subsystem and is available (and used by people) basically since day 1 on AM5.Reply
Why doesn't Asus do some real BIOS development and, for example, correct permanently broken MCR on 7000 CPUs instead of adding pseudo-features? -
The Historical Fidelity
Cause they’ve already sold their 600 series motherboards. They now want to sell their 800 series motherboards. They don’t care about your 7000 CPU problems because they already got your money.setx said:Setting FCLK to 2100+ is obvious tweak to boost memory subsystem and is available (and used by people) basically since day 1 on AM5.
Why doesn't Asus do some real BIOS development and, for example, correct permanently broken MCR on 7000 CPUs instead of adding pseudo-features? -
setx
I think 800 series motherboards have exactly the same problems as they are basically the same hardware.The Historical Fidelity said:Cause they’ve already sold their 600 series motherboards. They now want to sell their 800 series motherboards. -
The Historical Fidelity
It’s different hardware but yes gets the same job done. I wasn’t disputing that there are problems, I’m just emphasizing that Asus will only solve the problems that make their 800 series mobos more competitive to sell. They only care about getting your money, once they have it, they don’t care if there is more to be done when they have moved on to selling 900 series mobos.setx said:I think 800 series motherboards have exactly the same problems as they are basically the same hardware. -
rdgordon setx said:
"Why doesn't Asus do some real BIOS development and, for example, correct permanently broken MCR on 7000 CPUs instead of adding pseudo-features?"
You won't see much actual effort to deliver genuine performance upgrades to already released/sold hardware, as long as the media and outlets covering the industry just blindly glaze every empty "upgrade" released. They would much rather (both hardware vendors and the media who make money from affiliate links) see everyone buy brand new hardware every 18-24 months, even if the upgrades you're paying for could have been bios/software updates to your previous hardware.
For example, someone considering diving into AI/LLM for the first time, and don't know how they work, are who these updates/articles are aimed at. Anyone with half a clue understands that the bottlenecks for AI is typically GPU memory/throughput. If you're running a GPU with insufficient GDDR to hold the data needed to operate the model, the speed penalty applied to the compute is so huge using consumer grade, 2 memory channel processors. There are several generations of server/enterprise processors with 6 or 8 channels of DDR, and more PCIe lanes, that absolutely truck all consumer grade ryzen processors in AI/ML. And do it without doing things like decoupling the memory controller speed from the actual ddr speed. I have to laugh about this writer regurgitating the absolute slop about how turning off half the cores to the x3d CPU showing even better performance. This means 1 thing... the benchmark they're using is designed, or compiled, to use less than the maximum cpu threads available. Or they didn't overclock the CPU efficiently for the AI/ML load being tested, so it runs against a TDP or some other limit, and are benefiting from the lower overhead of only one CCD running, letting the remaining threads have a higher net wattage to work with.
Basically, a 2nd gen xeon scalable CPU with its 6 channels of ddr4 ram (heck, the higher performance gold xeon 62** processors are under $200 now, let's grab one of them so we can use sticks of optane dimms instead of standard ddr4 since they're $1 per GB now) will absolutely truck any brand new consumer grade CPU, even the zen5 99503xd, all else being equal (they also offer bifurcation so you can run 4 nvme as a single raid drive, doubling the write speed of pcie5 nvme drives, and you can run a bunch of those bifurcated drives.) -
bit_user
If you care about memory bandwidth, then you wouldn't use Optane DIMMs. They're like 1/10th as fast as DDR4.rdgordon said:Basically, a 2nd gen xeon scalable CPU with its 6 channels of ddr4 ram (heck, the higher performance gold xeon 62** processors are under $200 now, let's grab one of them so we can use sticks of optane dimms instead of standard ddr4 since they're $1 per GB now)
Anyway, at 2 DIMMs per channel, you could already fit 768 GB w/ 64 GB DIMMs, so I don't really know why you need Optane.
Getting back to memory bandwidth, the Xeon Platinum 8280 (flagship of the 2nd gen Scalables) supports only up to DDR4-2933. With 6 channels of that, your theoretical bandwidth is only 140.8 GB/s. That's not so much faster than today's gaming desktops. The equivalent DDR5 speed you'd need for a dual-channel setup would be DDR5-8800. -
Alex/AT Yeah, that's "AI" and as it's still giving bullshit, some minor errors from overclocking do not matter.Reply
Way to go.
( /s/ ) -
Makaveli
The 600 series motherboards and 800 series both use AMD’s Prom21 chip.The Historical Fidelity said:It’s different hardware but yes gets the same job done. I wasn’t disputing that there are problems, I’m just emphasizing that Asus will only solve the problems that make their 800 series mobos more competitive to sell. They only care about getting your money, once they have it, they don’t care if there is more to be done when they have moved on to selling 900 series mobos.
The only difference between X870E VS X670E is USB 4.
https://www.pugetsystems.com/labs/articles/amd-x870e-vs-x870-vs-x670e-vs-x670-vs-b650e-vs-b650/?srsltid=AfmBOop4bz6iB8f-92V692qm8lCth1fIPtRlonAHguhKe_AFk8qqogFc
https://i.postimg.cc/Gmq2vdQX/AM5-chipset.jpg