01:12 | GN Noise Chamber Up & Running
Check the video for this one!
03:21 | Rumor: Nvidia 40 Series Launch Dates Impending
First up, the GPU rumor mill has provided some alleged information on the upcoming Nvidia 40 series launch. Frequent leaker kopite7kimi on Twitter has “confirmed” a release order of RTX 4090 first, 4080 second, then 4070 third. The simple reply could also be read as an endorsement of launch dates seen in the 3DCenter.org tweet they were replying to. Taken at face value that would indicate launches in August, September, and October, respectively.
Igor Wallossek of Igor’sLAB guides a bit more conservatively in his article, speculating August or September will be the start of mass production, not full launch. He also provides additional context to the process timeline used in recent graphics card generations from Nvidia. Igor explains there are two distinct phases of board partner development: first “Learning,” then “Testing.” In the former, the board partner must understand new technological requirements, and in the latter must prove their designs. See the Igor’sLAB article for full detail.
Interestingly, Igor’s sources claim AD102 PCBs currently in testing are pin compatible with GA102, and that basic power and thermal testing is underway using GA102 packages with 600W vBIOS applied. This could mean that the 4090 will be using what are essentially 3090 Ti PCBs with minor tweaks, and that the 3090 Ti served double-duty as a proto-4090 test bed. This has been claimed a few times since the 3090 Ti launch and is starting to gain credibility.
Primary Source: https://twitter.com/kopite7kimi/status/1530972700910288896
Primary Source 2: https://www.igorslab.de/en/ad102-and-ad104-why-there-is-no-real-information-yet-about-nvidias-new-geforce-cards/
Secondary Source: https://videocardz.com/newz/rumor-nvidia-plans-to-launch-geforce-rtx-4090-in-august-rtx-4080-in-september-and-rtx-4070-in-october
07:00 | Rumor: Nvidia GTX 1630 Coming in June
The RTX 40 Series cards aren’t the only new cards coming soon from Nvidia. AMD also has the GTX 1630 to contend with -- and by ‘contend,’ we mean not that. In a leak posted by VideoCardz they claim a new partner embargo has gone out with a date of June 15 for general availability, but no media embargo is mentioned. Historically, NVIDIA hasn’t sampled low-end **30-series cards, like the 1030 or 730, to media for reviews; we’ve mostly bought them ourselves, and will likely do that again here. The GT 1030 and GT 730 have had a lot of shenanigans over the years, mostly relating to changing the memory supply, so we’ll keep an eye on the GTX 1630. Notably, it does have an ‘X’ in the name, not just ‘GT,’ so it’s likely to be perceived as comparatively higher end than previous X30 launches.
Previously, rumors pointed toward the GTX 1630 replacing the GTX 1050 Ti, a Pascal GPU from 2016. The 1050 Ti was a $140 card at launch (although there were a few versions) and, currently, NVIDIA doesn’t have a card in the $140 to $200 price range. The RTX 3050 comes closest, but its MSRP is $250. We’ll need to pay attention to the GTX 1630, but if it follows the general climb in prices against each rank in the GPU naming hierarchy, then $140 seems a likely spot to land. The XX30 cards have mostly been in the $100 and below price class previously, with the GT 1030 having an MSRP of $80. This would be a steep climb by name, but relative performance matters most.
In our most recent testing of the GTX 1050 Ti, the Pascal card ran at 96FPS AVG in this Rainbow Six Siege chart. That had the 1050 Ti as worse than the GTX 970, or compared to more recent cards, it allowed the RX 6500 XT a lead of 87%, with the GTX 1660 XC Ultra leading by 95%. Hopefully the GT 1630X runs a little faster. Current rumors suggest that the GTX 1630 will have less memory bandwidth than the 1050 Ti (despite a higher rumored memory speed), and that’s due to the 64-bit bus width. That said, delta color compression technologies have improved and the architecture has improved in its handling of memory, so there’s more to the story that we won’t know until we test the cards.
Primary Source: https://videocardz.com/newz/nvidia-geforce-gtx-1630-launch-shifts-to-june-15th-new-card-reportedly-slower-than-gtx-1050-ti
10:34 | Alleged Intel Raptor Lake CPU Benchmark Spotted
In a bit of Intel CPU-related news, another frequent leaker, TUM_APISAK on Twitter, posted a result from UserBenchmark alleged to be an engineering sample of the upcoming Raptor Lake architecture. The sample named “U3E1” reported 24 cores and 32 total threads.
We don’t like referring to UserBenchmark performance scores due to entirely uncontrolled test conditions, but in a situation like this, we can gain an extremely high-level indicator of the performance level on the sample. Versus the 12900K UserBenchmark results, the leaked sample posted 6% faster single-core results, for whatever that’s worth (nothing). We trudged through the overly Intel-biased landmines laid by UserBenchmark to check a Threadripper 3960X CPU, which also has 24 cores, and saw that UserBenchmark claims the new Intel CPU is “51%” better in average score. Upon looking at how that score is built, we saw that being 30 months newer is “hugely more recent” and therefore worth 100% more points.
Primary Sources: https://www.userbenchmark.com/UserRun/52976251 and https://cpu.userbenchmark.com/Compare/Intel-Core-i9-12900K-vs-Intel-0000/4118vsm1859895
Secondary Source: https://twitter.com/TUM_APISAK/status/1532349039567437825
12:33 | NZXT Z690 Motherboards Finally Launched
“What is up, fellow gamers!” is how NZXT chose to open its press release to announce the arrival of two new Z690 motherboards, perhaps a bit self aware of the fact these N7 motherboards are a bit late. For the naming, the “N” stands for “NZXT,” and the “7” stands for “7 months late after the launch of Alder Lake.”
The two new boards are the simply named N7 Z690 and N5 Z690, as usual for NZXT. The short names are actually somewhat refreshing after being bombarded with other extremely extreme and long motherboard names.
First off, the higher-tier board is the N7 at $299 US. The CPU Vcore VRM is a 12-phase design. It has four DDR4 slots, not DDR5, which is a choice we’re fine with considering that DDR5 is still maturing. Three M.2 slots for storage are present; the first two are PCIe 4.0 only while the third is either PCIe 4.0 or SATA. For PCIe there are three physical x16 slots (Gen 5 x16, Gen 4 x4, and Gen 3 x4) and two PCIe 3.0 x1 slots. WiFi and Bluetooth capabilities come via a small M.2 companion RF module in an additional E-Key “sicket”. Clear CMOS and BIOS flashback buttons are on the rear I/O, which are nice to have. CAM is tightly integrated for fan and RGB control. Visually, the N7 sports the usual NZXT look with flat white or black panels covering almost the entire board. The sections near the PCIe slots are held on with magnets and cover the M.2 slots and CMOS battery, which isn’t as buried as it has been in the past on NZXT’s motherboards. We are a little concerned about the chipset thermals under load considering this small heat sink is almost entirely covered by the cosmetic cover with no discernable contact between them.
The lower-tier N5 board comes in at $240 US. Differences from the N7 include 8-phase Vcore, four M.2 slots (including one strangely positioned on the rear), less USB, less analog audio outputs, no rear Clear CMOS button, and a somewhat better looking chipset heat sink. Visually there are less of the cover-plates resulting in a more modest but good looking board. One thing to point out is that the NZXT product page for the N5 lists “Optical S/PDIF” audio output on the rear I/O, while not being physically present on the images. NZXT should pay closer attention to details like this in case someone actually wants to buy one of these motherboards individually instead of it being forced on them through the BLD ecosystem.
Primary Sources: https://nzxt.com/news/explore-the-new-n7-n5-series-motherboards
17:32 | ASRock X670E and X670 Motherboards Announced
Back to the theme of new motherboards but this time with actually new chipsets. ASRock has announced five new socket AM5 motherboards for X670E and X670. Those are the:
- X670E Taichi Carrara
- X670E Taichi
- X670E Steel Legend
- X670E Pro RS
- X670 PG Lightning
The two Taichi motherboards seem functionally identical based on the little information currently on the product pages and the images themselves. The marble look of the Taichi Carrara is eye-catching and certainly stands out in the fields of endless black + RGB. They both feature huge 26-phase VRMs, two PCIe Gen5 slots (x16 and x8), and four M.2 slots. The M.2 includes the new “Blazing” M.2 Gen5 slot, which is definitely better than the “Hyper” ones. These are clearly high-end boards so expect lots of bells and whistles to be included.
The Pro RS motherboard is the only other with a picture so far, and appears to be a more modest affair with less flashy aesthetics, but it still has that Blazing M.2.
An interesting note is that all of the boards listed here have 2.5 Gb/s LAN ports, showing that multigig is continuing to catch on as a stopgap between 1 Gb and 10 Gb.
Primary Source: https://www.asrock.com/mb/index.asp#AM5
19:50 | HDPLex 250W/500W Tiny Passive PSU
We don’t often get into extreme small-form-factor products, but this one was too cool to pass up. HDPLex has launched a tiny PSU for ATX PCs that provides 250W of passively cooled power at 94% efficiency. HDPLex has had other more-powerful but still tiny ATX PSUs in the past by way of doing the AC to DC conversion in a separate “brick” unit, similar to how it’s done in laptops.
They boast that it’s the world’s smallest ATX PSU by using gallium nitride (GaN) FETs, an LLC+PFC structure, and a unibody flat transformer. Gallium nitride FETs were first seen in ATX PSUs in Corsair’s comparatively gigantic AX1600i. 250W admittedly doesn’t get you very far in today’s landscape of power-hungry silicon, so it’s possible to link two of them together for a combined 500W of power. HDPlex suggests using one to power the motherboard and CPU, while the other powers the GPU. This still might not sound like a lot, but the power density alone and brickless nature open up a lot of opportunities for the small-form-factor enthusiasts. This palm-sized PSU will cost $145 US and is expected to become available this month. No guarantees that we will test it, but we wanted to give this a highlight.
Primary Source: https://hdplex.com/hdplex-fanless-250w-gan-aio-atx-psu.html
21:21 | Thermal Grizzly LGA1700 Contact Frame
Next up, another overclocking oriented product from Thermal Grizzly and designed by extreme overclocker der8auer. The LGA1700 CPU Contact Frame claims to address issues brought on by the stock LGA1700 loading mechanism in conjunction with the elongated package Intel provided for Alder Lake. Thermal Grizzly explains it well here:
“The standard Integrated Loading Mechanism (ILM) has contact points that are in the middle of the elongated CPU. The surface of the Integrated Heatspreader (IHS) curves concavely due to the resulting uneven contact pressure of the processor in the socket. As a result, the base plate of the CPU cooler rests primarily on the edges of the IHS, so that the thermal ‘hotspot’ in the center of the CPU is not optimally covered.
“The Intel 12th Gen CPU Contact Frame has a special inner contour to shift the contact pressure from the center of the CPU to the edges during assembly. This avoids the concave curvature of the IHS. This means that the CPU cooler rests better on the processor and a larger contact surface is created to dissipate the waste heat of the CPU.”
Installation is shown in der8auer’s video for the Contact Frame, as well as the process of lapping the CPU for optimal flatness.
Shipping has begun to retailers so expect to see this in online stores soon. We actually have a few of these coming in and might test it if there’s a lot of interest, but can’t make any promises. We have some really cool pieces coming up so watch out for that.
Primary Sources: https://www.thermal-grizzly.com/en/products/523-cpu-contact-frame-for-12th-gen-intel-by-der8auer-en
23:06 | Lots of Steam Deck Competitors
Finally, Valve has now successfully bolstered an entire market segment of handheld gaming PCs with its Steam Deck, and so has new competitors from Aya Neo and AYN Tec.
AYN just launched its new “Loki” handheld, a portable PC running Windows OS and equipped with an AMD Zen3+ 6600U APU. This is one of AMD’s newer 6nm solutions. It’s accompanied by an M.2 SSD, USB4.0, and WiFi 6E, making it fairly robust in the I/O department.
The handheld has a set of colorful chassis options, including white, black, and custom vinyl stickers offering yellow, GameBoy purple, and NES or SNES-style coloring. We like how the chassis looks, actually. Face buttons are standard, there’s a D-pad on the left, and two analog joysticks are present -- no touchpad to be seen. AYN notes OS support for Ubuntu or Windows 10 and highlights that game streaming is an option for when the internal hardware is insufficient.
The AYN Loki also has an X-axis linear motor for rumble support, clearly taking a page from early Steam Deck criticism to offer an improvement.
We’d like to review one of these. If anyone from AYN Tec sees this, feel free to reach-out.
Aya Neo, whose products we’ve reviewed in the past, also has new stuff coming out.
Notably, there’s a new AYA Neo 2 Geek and AYA Neo Air Plus (with AMD Mendocino APUs). AYA Neo showed these in its Chinese-language 6-hour stream on YouTube, revealing that the Neo 2 Geek will run Ryzen 7 6800U parts and RDNA2 GPUs, with pricing landing at $800 USD maximally. The Aya Neo Air Plus will be $290, making it one of the cheaper handhelds out there. We’ll need to get these in for more analysis, though.
Secondary sources: https://www.tomshardware.com/news/aya-neo-2-geek-handheld-gets-ryzen-7-6800u-zen-3-and-rdna-2
AYA Neo Air Plus w/ Mendocino: https://www.tomshardware.com/news/aya-neo-plus-handheld-mendocino
AYA Neo 2 Geek with Ryzen 6000U Next Year
Host, Editorial: Steve Burke
Editorial: Jeremy Clayton
Video Production: Keegan Gallick