Review  When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works. AMD RX 9070 AI performance benchmark review vs 9070 XT, 7800 XT, Nvidia RTX 5070, 4070 Sayan Sen Neowin @ssc_combater007 · May 24, 2025 14:40 EDT Earlier this month, we shared the first part of our review of AMD's new RX 9070. It was about the gaming performance of the GPU, and we gave it a 7.5 out of 10. The 9070 XT, in contrast, received a full 10 out of 10. The main reason for the lower score on the non-XT was the relatively high price and thus the poorer value it offered compared to the XT. We thought the price was much closer to the XT than it needed to be. While the RX 9070 proved to be more power efficient than the XT, for a desktop gaming graphics card, value and performance typically take the front seat compared to something like power efficiency. However, that may not be the case in terms of productivity which also takes into account things like power savings. Thus, similar to the one we did for the XT model, we are doing a dedicated productivity review for the RX 9070 as well where we compare to against the 9070 XT, 7800 XT, as well as Nvidia's 5070 and 4070. AI performance is a very important metric in today's world and AMD also promised big improvements thanks to its underlying architectural improvements. We already had a taste of that with the XT model so now it's time to see how good the non-XT does here. Before we get underway, this is a collaboration between Sayan Sen (author), and Steven Parker who lent us their test PC for this review. Speaking of which, here are the specs of the test PC: Cooler Master MasterBox NR200P MAX ASRock Z790 PG-ITX/TB4 Intel Core i7-14700K with Thermal Grizzly Carbonaut Pad T-FORCE Delta RGB DDR5 (2x16GB) 7600MT/s CL36 (XMP Profile) 2TB Kingston Fury Renegade SSD Windows 11 24H2 (Build 26100.3194) Drivers used for the 7800 XT, 9070 XT and 9070 were Adrenaline v24.30.31.03 / 25.3.1 RC (press driver provided by AMD), and for the Nvidia RTX 5070 and 4070, GeForce v572.47 was used. (From the left) Sapphire Pulse 9070 XT, Nvidia 5070 FE, and Pulse 9070First up, we have Geekbench AI running on ONNX. The RTX 5070 gets beaten by both the 9070 XT and 9070 in quantized and single precision (FP32) performance. Similarly, the 4070 gets close to the 9070 in half-precision (FP16) performance, but the latter is an enormous 30% faster in quantized score and nearly 12.2% better in single precision (FP32). The reason for this beatdown is the amount of memory available to each card. The Nvidia GPUs have 12GB each and thus only do better in the FP16 precision tests since the other ones are more VRAM-intensive. Next up, we move to UL Procyon suite starting with the Image generation benchmark. We chose the Stable Diffusion XL FP16 test since this is the most intense workload available on Procyon suite. Similar to what we saw on Geekbench AI, the Nvidia GPUs to relatively better here as it is FP16 or half precision which means the used VRAM is lower. So this is something to keep in mind again, if you wish to float32 AI workloads, it is likely that graphics cards with greater than 12 GB buffers would emerge as victors. There is still a big improvement on the RX 9070 compared to the 7800 XT as we see a ~54% gain. This boost is due to improvements to the core architecture itself as VRAM capacities of both cards are the same at 16 Gigs. Following image generation, we move to the text generation benchmark. In this workload, we see the least impressive performance of the 9070 in terms of how much it improves over the 7800 XT. The former is up to ~7.25% faster here. The 9070 is also not as well-performing as the Nvidia 4070 in Phi and Mistral models, although it does do better in both the Llama tests. Another odd result stood out here where the 5070 underperformed all the cards including the 7800 XT in Llama 2. We ran each test three times and considered the best score and so we are not exactly sure what happened here. Wrapping up AI testing, we measured OpenCL throughput in Geekbench compute benchmark. The RX 9070 did not fare well here at all even falling behind the 7800 XT and it is significantly slower than the three other cards. Interestingly, even the RTX 5070 could not beat the 4070 on OpenCL so perhaps this suggests that OpenCL optimization has not been a priority for either AMD or Nvidia this time. It could also be an issue with Geekbench itself. Conclusion We reach the end of our productivity performance review of the 9070 and we have to say we are fairly impressed but there is also a slight bit of disappointment. It is clear that the 9070 as well as the 9070 XT really shine when inferencing precision is higher, and that is due to the higher memory buffers they possess compared to the Nvidia 5070. But on FP16, the Nvidia cards pull ahead. Still RNDA 4, including the RX 9070, see big boost over RDNA 3 (7800 XT). As we noted in the image generation benchmark, which is an intense load, there is over a 50% gain. So what do we make of the RX 9070 as a productivity hardware? We think it's a good card. If someone was looking for a GPU around $550 that can do both gaming and crunch through some AI tasks this is a good card to pick up especially if you are dealing with single precision situations or some other VRAM-intense tasks. And we already know it is efficient so there's that too. For those however looking for a GPU that can deal with more, AMD recently unveiled the Radeon AI PRO R9700 which is essentially a 32 GB refresh of the 9070 XT with some additional workstation-based optimizations. Considering everything, we rate AMD's RX 9070 a 9 out of 10 for its AI performance. Price is less of a factor for those looking at productivity cases compared to ones considering the GPU for gaming, and as such, we felt it did quite decent overall and can be especially handy if you need more than 12 GB. Purchase links: RX 9070 / XT (Amazon US) As an Amazon Associate we earn from qualifying purchases. Tags Report a problem with article Follow @NeowinFeed

Pros Sizzling hot throughput speedsExcellent 4K write speedWill be available in capacities up to 4TB Cons Lackluster PCMark 10 Overall Storage scoreLowest score on 3DMark Storage gaming benchmark of all the Gen 5 drives we've reviewed Lexar Professional NM1090 Pro PCIe 5.0 SSD Specs Bus Type PCI Express 5.0 Capacity (Tested) 4 Controller Maker Silicon Motion Interface (Computer Side) PCI Express Internal Form Factor M.2 Type-2280 Internal or External Internal NAND Type TLC NVMe Support Rated Maximum Sequential Read 14000 Rated Maximum Sequential Write 13000 Terabytes Written (TBW) Rating 2800 Warranty Length 5 All Specs The Lexar Professional NM1090 Pro (starts at $179.99 for 1TB; $494.99 for 4TB as tested) easily beats its rated sequential throughput speeds, according to our extensive tests, and it has the second-highest throughput scores we have seen from an SSD. But the NM1090 Pro's other test results are largely unimpressive; this PCI Express 5.0 internal SSD's sheer raw speed does not translate into great overall performance. So while it doesn't unseat the Editors' Choice-winning WD Black SN8100, it's an acceptable alternative. Design and Specs: Up to 4TB of Storage SpaceThe NM1090 is a four-lane solid-state drive running the NVMe 2.0 protocol over a PCIe 5.0 bus. This internal SSD comes in the standard M.2 Type-2280 "gumstick" format. This drive uses Micron 232-layer 3D TLC NAND flash and a Silicon Motion (SMI) SM2508 controller. Lexar cites this controller's 6nm architecture for helping to control the heat that Gen 5 drives are wont to generate and to minimize thermal throttlng. (Baffled by some of this lingo? Check out our handy guide to SSD jargon.)Physically speaking, this drive is also flashier than many of its competitors. It's adorned with a gold stripe across the front, which contrasts nicely with the black accents. It's sure to complement a classy PC build. The NM1090 Pro comes in 1TB, 2TB, and 4TB capacities. The 4TB stick that I reviewed is expected to go on sale in the US in the near future.Pricewise, the NM1090 Pro is midrange, below some of the elite Gen 5 sticks but higher than the modest-performing DRAM-less models we've tested: the Crucial P510 and the soon-to-be-reviewed Addlink G55H.As for durability, expressed as lifetime write capacity in total terabytes written (TBW), the NM1090 Pro matches the Crucial P510, T700, and T705 in the capacities they share. Its durability rating is a notch below the Corsair MP700 Pro, the ADATA Legend 970, and the Aorus 10000, which are rated at 700TBW for 1TB and 1,400TBW for 2TB. The Seagate FireCuda 540 is the reigning Gen 5 durability champ, with ratings of 1,000TBW for the 1TB stick and 2,000TBW for 2TB.The terabytes-written spec is a manufacturer's estimate of how much data can be written to a drive before some cells begin to fail and get taken out of service. Lexar warranties the NM1090 Pro for five years or until you hit the rated TBW figure in data writes, whichever comes first. But the drive's durability rating is such that unless you're writing unusually large amounts of data to the SSD, it's a safe bet that this drive will last the full warranty period and well beyond.The NM1090 includes the Lexar DiskMaster SSD Management tool, which lets you perform firmware updates, manage drive data, securely erase private files, and monitor drive health.System Requirements: Recent Motherboard RequiredPCIe 5.0 SSDs promise a major speed boost over PCIe 4.0 drives, but you can take advantage of it only if you have recent hardware that supports the standard. Only recent enthusiast-grade desktops and a handful of laptops are likely to be PCIe 5.0 SSD-ready off the shelf, so you may have to build your own PC from the ground up or update an existing system to gain the connectivity required. You'll need an Intel 12th Gen or later Core CPU with a motherboard based on Intel's Z690/Z790 or a more recent chipset; or an AMD Ryzen 7000 or 9000 processor with an AM5 motherboard built around an X670, X670E, B650E, or later chipset.Recommended by Our Editors(Credit: Joseph Maldonado)Now, an important point: Just because you have one of those chipsets doesn't guarantee that the motherboard maker actually implemented a PCIe 5.0-capable M.2 SSD slot. That's up to the board maker, so check your system's or motherboard's specs and documentation to make sure you actually have such a slot before investing in one of these drives. Some boards have PCIe 5.0 expansion slots for graphics cards and other PCI Express cards, but no PCIe 5.0 slots for SSDs. You need a PCIe 5.0-capable M.2 slot, specifically.Performance: Raw Speed Isn't EverythingIn benchmarking the NM1090 Pro, we used our latest testbed PC, designed specifically for benchmarking PCIe 5.0 M.2 SSDs. It is built around an ASRock X670E Taichi motherboard with an AMD X670 chipset, 32GB of DDR5 memory, one PCIe 5.0 x4 M.2 slot (with lanes that have direct access to the CPU), and three PCIe 4.0 slots. The system has an AMD Ryzen 9 7900 CPU using an AMD stock cooler; a GeForce RTX 2070 Super graphics card with 8GB of GDDR6 SDRAM; and a Thermaltake Toughpower GF1 Snow 750-watt power supply. The boot drive is an ADATA Legend 850 PCIe 4.0 SSD. (The reviewed SSD is tested as a secondary data drive.)We put the Lexar drive through our usual slate of internal solid-state drive benchmarks: Crystal DiskMark 6.0, UL's PCMark 10 Storage, and UL's 3DMark Storage benchmark. The last measures a drive's performance in a number of gaming-related load and launch tasks.Crystal DiskMark's sequential speed tests provide a traditional measure of drive throughput, simulating best-case, straight-line transfers of large files. We use this test largely to see if our tested speeds are in line with the manufacturer's rated speeds.The NM1090 Pro put up some scorching Crystal DiskMark numbers, easily beating its sequential read and write speed ratings while turning in the second-highest throughput results we've recorded, after the WD SN8100. And while its 4K read was middling for a Gen 5 drive and far behind the SN8100's prodigious score, the NM1090 Pro eked out a high score in 4K write, edging out the SN8100. Good 4K write performance is especially important for an SSD used as a boot drive, though we test them as secondary drives.The PCMark 10 Overall Storage test measures a drive's speed in performing a variety of routine tasks such as launching Windows, loading games and creative apps, and copying both small and large files.The NM1090 Pro's PCMark 10 Overall score was toward the low end of our comparison group, ahead of the two DRAM-less PCI Express 5.0 SSDs (the Addlink G55H and Crucial P510), as well as the two Gen 4 drives we included in the chart above. As for the individual traces that make up the PCMark 10 Overall score, the NM1090 Pro's scores were below average, generally similar to the DRAM-less PCIe 5.0 SSDs and ahead of the Gen 4 sticks. A bright spot was the NM1090 Pro's second-place finish in Adobe Premiere Pro loading (after the WD SN8100), though it was brought down to earth with a last-place finish in Photoshop loading, behind even the two PCIe 4.0 SSDs. The NM1090 Pro also had a last-place finish in the Windows loading trace.In 3DMark Storage, which aggregates a drive's performance at a variety of gaming-related tasks, the NM1090 Pro's score was at the bottom of the pack, with its nearest rival, the Crucial P510, besting it by 18%.

Menu Home News Hardware Gaming Mobile Finance Deals Reviews How To Wccftech Computex 2025Hardware ASRock Debuts New IPS, WOLED, And QD-OLED Gaming Monitors In The Phantom Gaming Lineup, 400Hz & 500Hz Options Sarfraz Khan • May 20, 2025 at 07:20am EDT The Phantom Gaming series gets three new gaming monitors in the mid-range and high-end categories, including a dual-mode monitor. ASRock Showcases Dual Mode 32-Inch WOLED Gaming Monitor, a QD-OLED 4K Monitor, and a 400Hz IPS QHD Monitor Almost every monitor manufacturer is debuting its dual-mode gaming monitor, and ASRock doesn't want to be left out either. The company is showcasing its latest innovations at the Computex event and has unveiled a few of its new offerings in the display department. ASRock displayed three gaming monitors in the Phantom Gaming lineup, introducing a WOLED dual-mode gaming monitor that can run at either 4K or 1080p resolution. The monitor model, PGO32UFS, is the company's first dual-mode gaming monitor, which it unveiled at the end of last year, and has brought it to the Computex table for hands-on. 2 of 9 The PGO32UFS monitor offers 4K@240Hz or FHD@480Hz mode on a WOLED flat display. The display size is 31.5" and has ultra-thin bezels on the sides for an immersive experience. The response time of 0.03ms GtG, the presence of AMD FreeSync Premium Pro, and various gaming-related technologies make it ideal for both immersive and fast-paced gaming. Plus, its accurate color reproduction with Delta E<2 color accuracy makes it ready for color grading as well. You will also find almost all the modern features on this one, including HDR compatibility, modern connectivity, a KVM switch for controlling multiple devices, an ergonomic stand for adjustments, and an integrated Wi-Fi 7 antenna for wireless internet connectivity. 2 of 9 The second one is also an enthusiast-grade gaming monitor, called the PGO32UFSA, which we learned about for the first time. ASRock hasn't yet prepared a dedicated page for this one, but has revealed its specifications and features at the event. The PGO32UFSA comes with a 32-inch screen size, boasting 4K resolution and up to 240Hz refresh rate. It's going to compete with the latest 4K@240Hz QD-OLED panels from manufacturers like ASUS, Gigabyte, and MSI, and has competitive features, as can be seen from the pic. 2 of 9 The third one is yet another enthusiast gaming monitor in the more affordable price segment, featuring an IPS panel that can run at up to 400Hz at 2K resolution. This is the PG27QFW2, a 27-inch gaming monitor that offers 1ms GtG response time, 93% DCI-P3 color gamut, and 400 nits of typical brightness. The monitor is decently ergonomic and also comes with dual 2W speakers for convenience. Subscribe to get an everyday digest of the latest technology news in your inbox Follow us on Topics Sections Company Some posts on wccftech.com may contain affiliate links. We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to amazon.com © 2025 WCCF TECH INC. 700 - 401 West Georgia Street, Vancouver, BC, Canada

TAIPEI—The Asus Back to the Future (BTF) initiative launched a few years ago in parallel with competitor MSI's largely similar Project Zero. These two ecosystems are all about PC aesthetics—or at least mostly are. The technologies employ what are known as “reverse connector” motherboards to make the inside of a desktop PC look as clean and as cable-free as possible. (A side effect of a cable-free interior is, potentially, cleaner airflow.)The Very Short History of Reverse-Connector MotherboardsReverse-connector boards, as name implies, have their header pins, cable sockets, and other connectivity on the opposite side of the board from usual. When you install one of these boards inside a PC case that supports these kind of boards, you connect all the wiring behind the board, in the cavity between the back of the motherboard and the right side of the PC case. (The right side panel is opaque, and hides all the cables.). And thus the interior, which is typically visible behind a transparent left side panel, looks great. These designs eliminate the tangle of power supply, case power, audio, and USB feeder cables. An Asus build using the new BTF 2.5 ecosystem, including an Astral RTX 5090 BTF Edition(Credit: John Burek)The one big cable that neither of the solutions could easily eliminate, however, is the big, gawky PCI Express power cable that runs in and out of a graphics card, which to an extent spoils the effect. A typical enthusiast PC will have a mainstream or better graphics card, and every one of these will have a prominent cable sticking out the top. Especially with Nvidia's GeForce RTX 40 and RTX 50 series, the 12VHPWR cable and connector (or the classic 6-pin or 8-pin PSU connectors) are impossible to miss and typically look like a giant blemish on otherwise clean build.Asus, with its first BTF designs, solved for that with the introduction of a handful of BTF Edition video cards, like this TUF Gaming RTX 4070 Ti Super. These custom cards were different than any others on the market in one very prominent respect: They have a power connector that looks like a second PCI Express edge connector on the bottom of the card, next to the big PCI Express data blade connector. This extra blade connector would interface with specially equipped BTF Asus motherboards to give the card power.The GPU power slot on an Asus BTF Edition motherboard; it accepts the power blade connector from a BTF Edition GPU.(Credit: John Burek)As you might expect, this solution created its own problems, or at least would long-term. Would most people be willing to buy an expensive BTF Edition graphics card that can't be installed on anything but another BTF board later on? Likewise, the selection of both BTF cards and BTF boards has been extremely limited to date. So there was something of a lock-in penalty for building with BTF Edition hardware and going all in. (MSI didn't have such a problem with its Project Zero because it didn't address the graphics card cabling at all.)We figured Asus or MSI would solve this issue in some clever way, eventually. It looks like Asus won the race.Here's a New Acronym to Know: GC-HPWRThe Asus answer doesn't do much to alleviate the limited selection of cards that can play this BTF game, but it does make those cards a much better future bet. At Computex 2025, Asus announced its BTF 2.5 initiative, spearheaded by two cards that employ a new power connector called GC-HPWR.GC-HPWR (short for "graphics card, high power") looks an awful lot like the underside edge connector on existing BTF graphics cards--but it's removable. When this connector is in place, you can only install the graphics card in a BTF motherboard. However, the GC-HPWR connector is on a module that you can wiggle and yank out of its socket. If you remove the module, the video card can then be installed in any motherboard that supports a GeForce RTX 50 series card, powered the normal, cabled way.(Credit: John Burek)So, you ask, how does the card get power if you pull out the GC-HPWR? The usual, familiar Nvidia 12HPWR power connector for a power supply cable is under a cap on the top edge of the card. Pop it off, plug in the usual Nvidia power supply cable or an adapter cable, and you're off to the races, like with any other graphics card.Recommended by Our EditorsYes, You'll Need a Special Card to Do All ThisAsus is rolling out two graphics cards with this modular GC-HPWR connector in two of its families: Astral (its recently introduced high-end enthusiast-class cards) and its stepped-down TUF Gaming. The first two cards announced at Computex 2025 are the TUF Gaming GeForce RTX 5070 Ti BTF White OC Edition, and the ROG Astral GeForce RTX 5090 BTF Edition.Asus did not share pricing for these two new swappable BTF cards. But it has now thrown down the gauntlet to MSI and any other players who may want to enter the cable-free desktop game.I got to handle the Astral version card at a pre-Computex event at Asus' headquarters in Taipei. The Astral card is one hefty beast, an RTX 5090 that you'll need a super-roomy chassis to accommodate. You can see it installed in a super-clean BTF 2.5 build here…(Credit: John Burek)Here you can see the underside of the card and the GC-HPWR connector. That's it at the bottom edge; if you look at the top edge, you can catch the outline of the cap covering the 12VHPWR socket. (Credit: John Burek)You can also see the cap here...(Credit: John Burek)The GC-HPWR may be modular, but it installs quite firmly. I was able to wiggle the connector loose with some effort. Here is what the module looks like removed from the GPU...(Credit: John Burek)As you can see on the top edge of the card, the flexible cap over the 12VHPWR is stealthy but easy to remove...(Credit: John Burek)In its initial presentation about BTF 2.5, an Asus representative pointed out that is theoretically possible to plug in both the 12VHPWR and the GC-HPWR at the same time, which would seem to be a very bad idea. They assured the audience, however, that plugging in both would not destroy the card and PC. (Still, we wouldn't test that.) Has Asus written the last word on cable clean PC building? We doubt it. This is a space that is ripe for standardization, the entry of new players in the market, and more vendor rivalries and drama. But Asus just laid down the law to rival MSI, and made it that much trickier for the two big remaining board makers, Gigabyte and ASRock, to come up with unique solutions that can compete without acceding to the other two board makers. We'll have to see where the cable-hiding wars head next: a peace treaty and a coming-together, or more battles?

Bryan from Tech Yes City tested Ryzen 7000/9000 processors on a few motherboards, revealing a difference in how ASRock motherboards supply SoC voltage in contrast to others. YouTuber Tech Yes City Demonstrates the Difference Between Motherboards from ASRock and Other Vendors; Analysis Indicates Both CPUs and ASRock Boards Cause Physical Damage If you are aware of the reports of dead Ryzen 9000 CPUs, particularly the Ryzen 7 9800X3D, nothing conclusive has been found regarding the issues that are causing these deaths. Perhaps this is one of the first investigations that shed some light on the issue and might help users understand what is actually causing CPU deaths. A few days ago, a popular tech YouTuber, Tech Yes City, reported his first dead Ryzen 9 9950X CPU on an ASRock X870 Steel Legend motherboard. This isn't the first time we have seen a Ryzen 9000 CPU dying on an ASRock motherboard, but there are actually nearly 200 such reports, most of them on Reddit. Bryan tried to find out what actually causes this, more on the ASRock motherboards than on boards from other vendors, and as per his investigation, there is a difference in how the ASRock motherboards handle the SoC voltage request from the CPU. SoC voltage is basically what the SoC section of the CPU needs for operation, and in most cases, it is static. It may fluctuate occasionally, but in the case of ASRock motherboards such as the ASRock X870E Taichi Lite with Ryzen 9800X3D installed, the SoC voltage kept fluctuating all the time. Even though the fluctuation isn't that significant, the upper limit is somewhat higher than what is considered the maximum limit. We can see that the SoC voltage in both cases (Ryzen 7 7700 and Ryzen 9800X3D installed) exceeds 1.250V and comes close to 1.270V. This is higher than what motherboards from other vendors could supply to the SoC of the CPU, and remains mostly near 1.20V, except for the ASUS X870E Crosshair Hero, which has by default added another 50 mV to add more stability. Nonetheless, it doesn't budge and remains constant all the time while the ASRock motherboard keeps fluctuating, which may result in permanent CPU damage, as we have seen before. Nonetheless, it should be kept in mind that it is the CPU that "dictates" how much SoC voltage is needed, and the deaths appear to be the result of how both CPU and motherboard handle the SoC request. Surely, this needs even deeper investigation as it isn't necessarily the only cause of CPU damage. Till then, ASRock might need to fix this through a BIOS update, but in case you want to mitigate the risk yourself, you will have to "enable" the Uncore Voltage from the BIOS. Deal of the Day