Available at https://www.bhphotovideo.com/c/product/1327199-REG/atomos_atomnjain1_7_1980x1200_4k60p_hdmi_2_0.html

Melbourne, Australia – 23rd March 2017: Today, Atomos brings the power of 4Kp60 10-bit ProRes recording and HDR monitoring to the GH5 for an amazingly low price of just $995.

“Our message to video Pro’s is to Go HDR 4Kp60 Today showcased by the GH5 and Ninja Inferno combo”, said Jeromy Young CEO and co-founder of Atomos. “For less than $3k it’s an unbeatable total package especially when considering having 4K HDR content ready for clients, Netflix and YouTube is a must”.

Atomos has always led the way advancing the quality, affordability and simplicity of filmmaking by adding professional features to popular Japanese cameras. Atomos are again first to launch a portable HDR 4Kp60 10-bit 422 monitor recorder. Timed with the release of the powerful Panasonic GH5 and a $995 price point Ninja Inferno empowers the masses.

Just as the original Atomos Ninja broke open DSLR filmmaking by giving the Canon 5DMKIII professional Apple ProRes recording & the original Atomos Shogun pioneered 4K with the Sony a7s and Panasonic GH4, the Ninja Inferno now arms the Panasonic GH5, the hottest camera of 2017, with HDR Apple Pro-Res 10-bit 4:2:2 4Kp60 over HDMI 2.0 – a feat not possible internally on the GH5. This marks another incredible Atomos breakthrough – professional 10-bit color resolution, 4:2:2 color accuracy, high frame rate 4K 60p video resolution & all with the incredible brightness range that HDR delivers in PQ or HLG.

Apart from being an obvious companion for the new GH5, the Ninja Inferno is the world’s first HDMI monitor-recorder to accept 4K DCI signals from cameras like the Panasonic GH4 / DVX200 / HCX1000 / UX180 / HCX1, Sony FS7 / Z100 and the JVC LS300. Support for the 4096 DCI standard unlocks cinema recording from these 4K video & mirrorless DSLRs.

For the Ninja Inferno, like its flagship sibling the Shogun Inferno, we have created the ultimate monitor through end-to-end custom engineering of all components. It starts with a 1920x1200 resolution LCD panel and add 10-bit processing in highlights and equivalent 10+ bit resolution in blacks. This is achieved through the patented AtomHDR engine which when combined with our custom Atomos-built backlight allows the power of 1500nits to be utilized for High Bright Rec709 or HDR PQ/HLG at 10+ stops. The full-size HDMI 2.0 connection bypasses the camera’s internal limitations by recording to 4Kp60 4:2:2 10-bit pristine video in grading-friendly Apple ProRes or Avid DNxHR. The Ninja Inferno upgrades mirrorless and DSLR camera audio by including balanced XLR input and 48V Phantom Power, negating the need to purchase separate audio recorders. For on set review, the wide angle 7” calibrated screen, playback controls, playlist and XML tagging make the Ninja Inferno the perfect client or director’s monitor, with playout to the latest HDR PQ/HLG or SDR TV’s. The Ninja Inferno is equally adept in the editing suite as it is in the field with the recorded files dropping directly onto the timeline of all the major NLE editing and grading software with the HDMI input capable of accepting HDR signals to make the Ninja Inferno an unbelievably affordable HDR reference monitor for portable or in studio color grading.

The Ninja Inferno, which is shipping now for MSRP $995, is the ultimate camera accessory for the GH5, the popular Sony FS7 or cameras with 4K DCI output & other 4Kp60 / 4Kp30 / HDp60 cameras looking for a future proof monitor-recorder.

HDR now comes free across the Atomos 4K Line-up

Timed with the release of the Ninja Inferno, Atomos have driven down the price of the existing line-up, passing on the volume savings they receive due to the popularity of the entire HDR range. The Shogun Flame and Ninja Flame are now $995 and $795 respectively (both models ship with mains power supply, recording caddy & travel case. The Shogun Flame also ships with an XLR breakout cable). To complement all models in the line-up, Atomos have developed a tailored Accessory Kit and Power Kit that arm users with everything needed to power, control, dock, charge and safely carry the units at an amazing price. The flagship Shogun Inferno continues to ship as is today.

https://www.ebay.com/itm/MNPA19-XTR-10GB-MELLANOX-CONNECTX-2-PCIe-X8-10Gbe-SFP-NETWORK-CARD/131634470127

$17 adapters, note that they require are PCIe v2.0 x8, so can use 8 lines and this could be quite bad, as on many MB's second PCI-E x16 slots are x4 only. Yet such single port adapters work perfect in x16 slots where only 4 lines are available in reality. So, do not worry about this.

You also need direct SFP+ DAC cable, it is plenty of such cables at 1-3m, but 5m cables are much more rare and expensive.

https://www.ebay.com/itm/SFP-H10GB-CU3M-Cisco-10GB-SFP-Twinax-Copper-3M-Passive-DAC-Cable/273204097845?

If you need to go longer than 5m you will need optical patch cord

So, need pair of these (850nm 300m SFP+), they are around $16 each

https://www.aliexpress.com/item/6COM-Compatible-for-MikroTik-S-85DLC03D-10Gb-SFP-module-10GBase-SR-MMF-850nm-300m/32845294561.html

And after this need to get optical duplex multi-mode patch itself, note that they can be OM2, OM3 and OM4

• OM2 has 82m maximum distance for 10GBASE-SR
• OM3 has 300m maximum distance for 10GBASE-SR
• OM4 has 400m maximum distance for 10GBASE-SR

So, in out local home reality anyone of them will do.

Note that both optical patch cord and SFP modules

https://www.aliexpress.com/item/10M-LC-LC-DUPLEX-10-GIGABIT-50-125-MULTIMODE-FIBER-OPTIC-CABLE-OM3-AQUA-10GB-PATCH/32567430719.html

Info

• 12.5" model is $525 Core M3/ 4GB RAM/ 128GB SSD/ USB 3.0 and HDMI ports + USB-C port, 11.5 hours battery
• 13.3" model is $750, Core i5-6200U/GeForce 940MX/8GB RAM/256GB SSD, 40Wh battery, 0.6" thick, 2.8 pounds

Here a little explanation what I do: We film on a GH1 (Hacked), GH3 and GH5. The GH5 is used in 4k 10 bit. Output is HD. The shots of the GH5 are quite often used to mimic more cameras (crop in) and to mimic a slider.

Up to now I use a Lenovo W541 (i7-4910MQ, 16GB RAM, Quadro K2100M) All data on internal SSD (Ultrabay) seperate from OS). It is really hard to work with this PC and the 4k material of the GH5.

I am thinking of upgrading to an X1 Extreme (i7-8750H, 16GB RAM, GeForce GTX 1050 Ti 4 GB, M.2 SSD)

I want to stick with Lenovo. We use quite special Multichannel Audio interfaces (RME Madiface XT). We had some problems with other brands. Since years we stick to Lenovo and never had problems. So this will not change. I am not interested to get a seperate Laptop just for video work.

The question is if the update will speed up things considerable?

Thanks for your input about this.

Best. Fabian

On the back it looks like

Issue with it is that cooler mount is just 4 bare holes in motherboard - totally unacceptable solution for anything besides stock cooler if you tray to use them as they are. As motherboard will bend.

So any good cooler manufacturer ship you their own extra backplate and mounting hardware.

Typical accessories set

This should not be the case. each motherboard must have proper backplate suitable for any cooler.

This thing can easy take two 360mm radiators in addition to your HEDT build with multiple GPUs :-)

And also good for cooling back of VRM due to special air guide.

In EU they are cheaper, btw.

Personally recently saw such build and with good fans it is very silent with outstanding airflow.

Look at PCIe connectors pinout, it is very interesting thingy.

As its origins can tell you a lot about committee, bureaucrats who are present in them and highly specialized engineers who are forced to go around stupid decisions.

Bureaucrats reversed wires assignment compared to ATX CPU connector, in absolutely stupid attempt to provide reason why they must be paid for their work.

During design bureaucrats realized that many capitalist manufacturers will put crap thin wires in their PSU for PCIE power delivery. Proper sane man will work with ATX standard and set strict specifications to always use 3 wires for 12V 6 pin connector (and 4 for 8 pin one) and 3 wires for ground, copper ones and no smaller than specified gauge. But no.

Decision looked smart for bureaucrats but in reality it is utter stupid. They added special sense wire so properly designed PSU will sense actual voltage drop across the crappy wires and adjust +12V going only to this connector to compensate for this drop.

Well, at first. It is cheap crappy power supplies that save each cent and use thin alu wires. They won't do extremely advanced implementation you want from them. And expensive power suppliers with good thick wires do not need it much.

Second. It requires separate step up DC-DC regulator for each PCIE connector inside PSU. It is around 20x more expensive than to use good wires. I am not even talking that none of standard IC for PSUs have support for such features or that it is not much space inside PSU for all this shit.

Third. GPU manufacturers are also not stupid and know that they can't rely on top quality +12V values at mass market cheap PSU, so they try to design their power stages with good margin or voltage drop.

Perfect example of people who got lot of money, wasted lot of time and instead of simple copy of ATX CPU connector induced problems no one asked for.

It is all something like this one Youtube sketch, just in real task:

• 10.5 inches AMOLED 1600 x 2560 pixels, 16:10 ratio screen
• Qualcomm SDM670 Snapdragon 670
• 4GB RAM+64GB NAND
• MicroSD support up to 512GB
• 13Mp main camera + 8Mp front one
• USB Type C 3.1 port
• Has adapter for headphones
• Wi-Fi 802.11 a/b/g/n/ac 2.4Ghz+5Ghz, Bluetooth v5.0
• Android 9.0
• UHD 4K (3840x2160) @ 30fps video record
• Can play UHD 4K (3840x2160) @ 60fps in HEVC and H.264
• 7 040mAh battery
• 4 speakers
• 245 x 160 x 5,5mm
• 400g
• $399

So, what models to look for on used market if you plan to use it in video editing?

First. Most frequent used cards in mining are - AMD RX570 (and RX470) 4GB/8GB, AMD RX580 (and RX480) 8GB, NVidia GTX 1060 3GB, NVidia GTX 1060 6GB.

All of them are very good in editing.

Note! Mining do not do anything horrible with the cards if rig had been maintained properly, do not fall for horror stories.

Second. Issues to look for:

Fans. Make sure to test fans at all speeds, at cold state and after warming. Lot of modern GPU fans are not too good after prolonged work.

You can also look for card with silent fan mode, where fan stops under low load (it is around 80-90% of editing work time).

Thermal pad leaks. Famous issue of some cards, looks like oily areas, or just slightly oily looking board parts

Contrary to this photo, it is usually located under cooler close to RAM or VRM area, and you can miss some of it. Such thing appear due to vertical position of the card in most mining rigs contrary to horizontal position in almost every computer case. Thing that is leaking is dielectric, so nothing bad must be happening.

If card works good, and discount is big, plus you are able to replace thermal pads you can buy such card.

Card under load. Use https://geeks3d.com/furmark/ and https://benchmark.unigine.com/superposition to check card under load for at least 30 minutes, it must not be crashes or visual artifacts, same as abnormally high temperatures. Can also check card model using https://www.techpowerup.com/gpuz/

Third. Cards to prefer - AMD RX580 8GB, NVidia GTX 1060 6GB, GTX 1070 Ti 8GB, GTX 1080 Ti 11Gb

Try to buy more powerful card with more memory. Video software, especially Davinci Resolve or advanced plugins is GPU memory hungry and become more so with each new release.

Mac/Hackintosh guys. Many Mac users buy AMD cards as external ones with special Thunderbolt case, yet it has lot of issues, check good articles and videos first. It is still way cheaper than pay for BM external solutions.

GTX 10xx series had good drivers and supported by recent MacOS versions (but Nvidia Drivers for Mojave 10.14 Not Available!). Just make sure to check Hackintosh special resources before doing anything.

If you are on super tight budged - your choices are good 1050 Ti 4GB models, GTX 1060 3GB and RX570 4GB (can be usually found for $120-150 used).

Cooling to prefer

Avoid so called turbine coolers on cards

Lower end cards are fine with just good large two coolers.

Best under moderate and low load are usually three cooler long cards (as they have bigger dissipation area and lower coolers speed)

Can check ASUS Strix line for such.

Fourth. Warranty. Many cards still have 2-3 year of warrant left, so you can buy such for more safety. In many countries outside US/EU you can see cheaper cards with no local warranty at all as they had been shipped from US/EU. If discount is big and seller provide his own small warranty you can try to risk.

Last. Newest RTX 20xx and AMD RX 590 cards provide very little for video editing and is considered non optimal generation due to little progress and price hikes. Most advantage is in very heavy tasks and usage of RTX 2080 Ti. So, for almost anything, including advanced grading you will be good using older card.

Note. Usually local sellers provide best deals and it is possible to check card, contrary to ebay. Check your country most popular C2C platform to buy used stuff.

http://www.intel.com/content/www/us/en/nuc/nuc-kit-d54250wyk.html

NUC. Now on Haswell :-)

BRIX. Same shit.

Main marketing push is efficiency. Yet beyond good Bronze PSU it is questionable, beyond Gold it is stupid.

Rising efficiency means that you need to reduce losses in elements of PSU.

Bronze PSU do this by going from group stabilization to DC-DC converters for 5V and 3.3V. Good Bronze PSU can have very high efficiency and use elements (mostly MOSFETS) designed for much higher load to further reduce losses.

Gold PSU is doing efficiency improvement mostly by switching to so called LLC resonant half-bridge scheme (some PSUs even use full-bridge to gain 1-2% more efficiency).

Further progress requires to put even more beefy elements and ... actually reduce voltage quality.

Main savings start to come from reducing number of transistors switching events. The lower the PWM frequency - the higher the ripple voltage (you can compensate it with capacitors, but still).

It is much better idea to put nice good heatsinks (very cheap extruded alu) and high quality fan and reduce it's speed a lot where possible.

But in reality we have huge marketing expenses, certification costs, stupid idea of half/full passive modes, Japanese capacitors craze and much more of such shit.

QLC line looks nice, isn't it?

This is example of capitalist greed, each of them want maximum profits no matter the damage made.

3D QLC production cost is slightly higher compared to 3D TLC, to compensate for it and further increase profits lot of firms simplify chip designs in logic part. Making it less channels and use simplified design.

Present profits of firms like Samsung and Micron selling latest 65-96 layers QLC and TLC can reach as high as 500-2000% from production costs. Similar to drug cartels and their methods are exactly the same - bribes and corruption.

For almost 3 years we have cartel agreement that only allow moderate slides after strong push (this time it had been Chinese agencies, rumors are that they told that otherwise they make public real cost calculations).

NVMe controllers profit margins are also huge, controller manufacturing cost is usually LOWER than SATA one due to simpler protocol, yet wholesale chip price can reach 5x-10x of SATA controller prices, and all major manufacturers also meet regularly and set coordinated prices.

Next layer of protection is manufacturers of equipment used to make NAND chips, they are tightly controlled, and are part of cartel agreement. Their interest - keep huge margins on their equipment and do not allow any competition.

Same goes for Optane memory. Intel wants to control production volume and tightly control speed and prices (hence their conflict with Micron). Same Intel is in constant talk to NAND manufacturers.

Industry rumors are that current Optane memory is around 4 times slowed down, both by design and controller. As well as prices are kept artificially 4x higher than they could be. Intel goal is to go slow and not destroy NAND industry, as well as being able to get huge profits at corporate and datacenters areas.

Capitalism forces firms and people to hide information where public is fed with fairy tales. And it also huge impact on real progress.

First sad thing

Every company try to advertise non existing numbers.

For example, for good USB adapter it can be AC1900 or 600Mpbs on 2.4GHz band and 1300Mbps on 5GHz band. At first, forget 2.4Ghz, in any real place it'll be so crowded that almost any good adapter or router will have similar low speeds. And you can't sum the two anyway.

1300Mbps here means that adapter is 3x3 MIMO, or has 3 separate antennas used both for transmission and receiving of signals each in separate band.

• 1x1 MIMO in all cheap adapters can theoretically reach 433Mbps.
• 2x2 MIMO in more advanced notes and most routers can theoretically reach 867Mbps.
• 3×3 MIMO can be found in lot of routers/access points and also in really good notes/smartphones (like Apple and Samsung).
• 4x4 MIMO in top routers and few top adapters can theoretically reach 1732Mbps.

First Note! You need really almost no interference for all this to work good. Otherwise speed will start to drop really fast from limits mentioned.

Second Note! Speeds mentioned are bare theoretical speeds including all correction codes, protocols expanses, etc and have very small relation to real transfer speed you can measure.

Good case test (2x2-4x4 access points connected to 3x3 Mac Pro)

Our adapters choice - used Dell Broadcom 57810S Dual Port 10GBASE-T Converged Network Adapter

For $45 only per adapter (cost of one good new single port Intel 1Gbit separate card)

https://www.ebay.com/itm/Dell-Broadcom-57810S-Dual-Port-10GBASE-T-Converged-Network-Adapter-HN10N-Low-P/132761054157

Few main reasons:

• Cheap
• Ability to attach adapter to different networks or join forces of two ports to reach 20Gbit (but usually it'll be only for multiple transfers).
• Official support for RDMA, rare for adapters outside Mellanox. Note that actual support for it in Windows drivers must be checked. I saw negative reports on this.

Stay away from relatively cheap (<$110) new copper based 10GBASE-T adapters, usually using Aquantia AQtion controller. They are fine things, but made for consumer market and do not have server grade drivers with necessary features and not designed for low CPU utilization.

On cables

If length of the cables is small you can use worse cables than recommended, here I mean old Cat 5e patches and cables, in case they are good quality and not damaged. Plus EMI in your place is not extreme. Cat 6 cables just have more turns in each twisted pair per same length, plus use thicker wires. Chinese Cat 7 patch cords are also pretty nice shielded things, if not too long (as they sometimes can save on copper :-) )

Otherwise follow official recommendations and in case of significant EMI use shielded cables.

That is pure modular thing, made to perform certain function.

To dreams of idiotic marketing:

And it is not just fun, as for you pay around 2.5x the possible normal price for board with good quality and VRM.
All this extra 150% go into bells and whistles, marketing campaigns, paid and just free prostitute influencers and test boards for them.
And you also pay for all integrated features, even if you don't need some. Each time, for each new board.

In history we have visual parallels, as process going in society is also very similar

Sans-culottes

Muscadins - les Incroyables et Merveilleuses - sign of Thermidorian Reaction in France

We assume that network will be made for individual or small team and will be made from ground up, not using existing cables and such.

Specific topics:

• https://www.personal-view.com/talks/discussion/21364/making-10gbit-network-using-sfp-adapters-for-cheap#Item_14
• https://www.personal-view.com/talks/discussion/21443/going-further-40gb-and-56gb-infiniband-for-cheap-mellanox-and-windows-10#Item_3

Most i7 and i9 modern platforms still do not have HDMI 2.0 ports... because of HDCP, LSPCON ships are expensive if they have proper implementation of HDCP 2.2.

Lot of older TVs have HDMI 2.0 ports, but not with HDCP 2.2 support disabling output if you watch legal 4K content and making you learn useful stuff to watch same content in higher quality and much cheaper :-)

Of course Intel famous laziness, as they did not change their GPU part for 3-4 generations, also plays some role here.

Blocks was made in software in multiple places like CPU microcode, iGPU UEFI GOP, Management Engine Modules in BIOS.

• http://www.overclock.net/forum/225-hardware-news/1665657-coffeelake-completely-working-100-200-series-chipsets.html
• https://www.win-raid.com/t3413f16-GUIDE-Coffee-Lake-CPUs-on-Skylake-and-Kaby-Lake-motherboards.html

Of course if power part of MB is weak it won't help with running 6 core CPUs.

If core value is good looks and only transaction moment instead of fulfilling needs - this is that you get. And RGB, of course.

And indeed, in spec sheets you see it in A320, B350, X370, B450 and X470.

But, in reality in 95-99% of all cases you won't find them used on motherboards.

Instead motherboards makers opt out for additional controllers from... ASMedia - 1142, 1143, 2142 and 3142.

And even more fun fact - two of them are significantly worse than build in controller, and two others are slightly worse and also take two PCI-E lines that could be used for other things.

Implementation using build in controller need recklockers made by... ASMedia and incidentally they had been pricey and in very short supply constantly.

So, Asmedia implemented special scheme making each and every board around $3-7 more expensive and lacking some other feature due to this and also due to less space on board.

This is how capitalism works. Induce billions in damages to make your own little income stream.

Focused on small teams and individual video editors needs.

I'll be adding systematic info post by post.

Ideal modern storage ITX motherboard must allow:

• Work with stock or slightly overclocked 8 core CPUs fine without VRM issues, in case you will want to use one for distributed rendering and will use RDP.
• Usual CPU it will work with will be 2200G or 2400G 4 core with integrated GPU, if you want to use it for some previews and backup editing machine.
• Ideally must allow to install pair of NVMe x4 SSD drives (one can be using PCIe 2.0).
• The more SATA 3.0 ports - the better, support for SATA ports multiplier boards with ASMedia chipset is nice.
• Have pair of USB 3.1 Gen2 10Mbit ports so you can transfer data from/to your fast SSD quick - check more
• PCIe x16 slot will be taken by 10Gbit/Infiniband adapters to quickly use directly or dump large amount of footage at NAS/Server, so if some board have integrated 10Gbit adapter - it is also huge plus for expandability. But now such are absent on the market.
• In case of distributed rendering you need quite good external GPU and revert to use 1Gbit controller

CPU info and limits

Ryzen is gaming/consumer CPU with some IO shortcomings by design, not so important for ITX boards.

• Ryzen chips have 24 PCIe v3.0 lines.
• 4x goes to NVMe slot of board
• 16x go to GPU - and this one on A320 and B350/450 can't be split (aka bifurcation), to force MB makers to buy more expensive chipsets for SLI.
• 4x is left for the chipset. Chipset uses them for lot of things, including making multiple PCIe 2.0 lines.

Note! Ryzen 2200G or 2400G are designed such way that they provide only 8 PCIe 3.0 lines to GPU instead of 16x and they can't be split. All else is same. For ITX case it is not so important.

Chipset difference and motherboard limits

For ITX boards difference is extremely small, if exist, as usually it is only one PCIe slot, and bifurcation can be actually enabled on some boards even with B350/B450 chipsets and disabled on others using X370/X470 one.

Note that some motherboard can refuse to boot without GPU installed (if you use CPU without such) and this also is BIOS version dependent.
I'll try to add this info later.

You are looking for Mellanox QDR (and FDR) adapters, preferably double port. Used, of course.

They come from supercomputers and datacenters focused on high speed storage and calculations.

Check about Infiniband kinds at https://en.wikipedia.org/wiki/InfiniBand

Cards must have QSFP+ ports, FDR are more expensive (and faster).

Mellanox ConnectX-3 VPI CX353A-FDR InfiniBand 56GbE/40GbE for $55

Mellanox ConnectX-3 MCX354A-QCBT CX354A QDR 40GbE Dual Port for $42.5

In reality last letter is bracket height - S (Short Bracket) / T(Full-Height Bracket).

http://www.mellanox.com/related-docs/user_manuals/ConnectX-3_VPI_Single_and_Dual_QSFP_Port_Adapter_Card_User_Manual.pdf

You will also need proper cable, for short distances it looks like (it is same direct soldered wires as in SFP+ case for Ethernet adapters, just more of them)

https://www.ebay.com/itm/Original-Mellanox-HP-3M-4X-DDR-QDR-QSFP-IB-Cu-Cable-498385-B23-503815-002/113404887644

For long length you need optics:

https://www.ebay.com/itm/Mellanox-MC2207310-010-10m-Infiniband-Optical-cable-QSFP-QDR-FDR10-FDR56-rate/183270585614

On software side

First, you need to check drivers availability (we'll assume that you will use Windows 10), usually it is not problem for popular Mellanox cards.

Usually rebranded cards can also work fine with Mellanox drivers.

Second, you need proper software - Windows 10 Pro for Workstations specifically, with properly enabled [RDMA] (https://en.wikipedia.org/wiki/Remote_direct_memory_access) and SMB Direct.

Without this options you will be mostly wasting your time and money (can check Linus making same error and not understanding it).

With RDMA

Without

Read about them at:

• http://wchukov.blogspot.com/2018/05/infinibandrdma-on-windows.html
• https://www.servethehome.com/custom-firmware-mellanox-oem-infiniband-rdma-windows-server-2012/

Check second link also if you need to make custom firmware (to enable RDMA!) in case your adapter is rebranded one.

Another big thing with Infiniband - it is solution to go if you plan to use Intel Optane memory as main editing drive.
With it you can do it remotely from other machine and still get amazing IOPS and keep low latency.

Using PCIe 4.0 X16 Adapter - with work with all new Ryzens and Threadripper CPUs.

http://www.mellanox.com/page/products_dyn?product_family=268&mtag=connectx_6_en_ic

https://www.hardwareluxx.de/community/f12/pga-am4-mainboard-vrm-liste-1155146.html

Real authors of USB-C is the smartphone manufacturers, who wanted reversible port like Apple have and also wanted USB 3.0 speeds without bulky long and prone to breakup USB 3.0 micro port (it is really awful).

Yet, USB-C port is expensive and fragile for industry outside small gadgets. Yet media, especially ASMedia :-), want you to believe that you really need USB-C for fast 10Gbit transfers, it is not so.

In reality in each orientation only two pairs are used (one RX and one TX).

As we can see at USB 3.0 Type A connector pinout - it also has same lines for USB signals (hence Gen 2 motherboards and adapter using them).

Now to USB 3.2 20Gb mode (aka 2×2)

USB 3.2 takes advantage of the four differential SuperSpeed/SuperSpeedPlus pairs present in the USB Type-C connector, unlike USB 3.1 and USB 3.0, which used one or the other TX/RX lane pair, depending on Type-C connector orientation.

This will require to have USB-C port.

A USB 3.2 implementation achieves a 20Gbps raw data rate by lane striping and lane bonding (e.g., splitting and combining data) with two USB 3.1 (10G) lanes. USB 3.2 also supports 10Gbps by striping and bonding two USB 3.0 (5G) lanes. USB 3.2 supports USB Type-C features like Alternate Modes, Power Delivery, and Digital Audio.

The USB-IF is emphasizing the transition to Type-C by moving the USB cable and connector chapter to a separate document and renaming the standard-A, standard-B, and mini/micro connectors as legacy USB connectors.

All passive USB Type-C cables can be used for USB 3.2 GenXx2 connections since four SuperSpeed/SuperSpeedPlus differential pairs are mandatory per the USB Type-C specification. A passive cable designed for Gen2 (10G) is limited to approx. 1m length and can support the new 20G connection speed. Two- to three-meter passive cables designed for Gen1 (5G) can support the new 10G connection speed.

Melbourne, Australia – 17th April 2016: Atomos invented the all-in-one monitor recorder category and now redefine it with the Shogun Inferno. It is the combination of the most advanced monitor technologies – 1500nit/10bit/HDR, the latest recording capability – 4Kp60 over Quad-SDI, along with playback & editing functionality that encourages on set collaboration. Shogun Inferno makes mastering a high resolution, high frame rate, high dynamic range workflow available to all filmmakers.

Shogun Inferno sits as the flagship model in the new 8-strong Atomos HDR line-up which has expanded to include HDR as a free update to the Blade HD series, Ninja Assassin, Shogun and Shogun Studio along with the recently announced Flame series.

Shogun Inferno has all of the premium monitoring features previously announced on the Shogun Flame i.e. 10 stops of dynamic range with AtomHDR, 1500nit brightness for outdoor monitoring & 10-bit processing for smooth color transitions.

• AtomHDR; AtomHDR lets you shoot with the high brightness range of your camera’s Log profile and previewthe final, vibrant post production HDR result.
• 4K 60p; Record and play out pristine ProRes/ DNxHR in 4K resolution and 24/25/30/50/60p frame rates.
• HD 240p; Apart from 4K recording, high frame rate HD from 50p to 240p can be recorded from cameras thatoutput these high frame rates.
• Quad SDI; The Quad SDI inputs let you connect from cameras with 1.5/3/6/12G SDI outputs without the needfor converters.
• Raw recording; Capture the RAW output from Sony FS7/ FS700 and Canon C300MKII/C500 over SDI,recording to either ProRes, DNxHR or CDNG.
• 10-bit monitor; 10-bit monitor processing increases the number of colors from 16.7 million for standard 8-bitpanels to 1.07 billion, minimizing color banding on screen.
• 7” 1500nit brightness; For Standard Dynamic Range (SDR) shooting ramp the brightness slider up to 1500 nitfor hood-free outdoor monitoring.
• Apple ProRes & AVID DNxHR recording; Record to visually lossless Apple ProRes or AVID DNxHD in Rec709or Log formats as an edit friendly, visually lossless industry benchmark independent of the camera brand used.
• PQ in/out; Take PQ out into larger HDR compatible screens or feed HDR from your NLE into the PQ input forHDR grading using the Inferno.
• Custom looks; Apply a custom look to footage by uploading and applying “.cube” 3D LUT’s. View in full/halfmode on screen, output it to a monitor or record into the footage.
• Continuous power; Our patent pending Continuous power system automatically swaps to the second batterywhen power is low for uninterrupted recording in the field.
• Playlist generation; Create playlists easily, either entire clips or tagged parts of clips, for playback on the unit orout to a larger screen.
• High quality 4K 60p and HD 240p ProRes and DNxHR recording, from either a video or Raw signal – perfect for sports broadcasters and slow motion recordings.

The Shogun Inferno ships with host of accessories including a HPRC hard case, a snap-fast HDR sunhood, fast battery charger, control cable, 2 x NP-F750 4 cell batteries, USB 3 Docking Station, D-Tap Cable, 5 x Master Caddy Cases, Break-out XLR cable and a 12 V 3A AC-DC Power Adapter. MSRP is US $1995 with availability in Q3 2016.

The Shogun Inferno is shipping Q3 2016 for a MSRP of US: $ 1,995 ex. TAX EU: € 1,995 ex. VAT GBP: £ 1665 ex. VAT AUD: $ 3099 Inc. GST JPY: Yen 278,000 inc VAT

https://proav.roland.com/global/products/vr-1hd/?lang=en-JM

Long ago it must become similar to USB 3.0 internal connector