enfrdepluk
Search find 4120  disqus socia  tg2 f2 lin2 in2 X icon 3 y2  p2 tik steam2

Reviews NVIDIA GeForce RTX 3080 Founders Edition

Today NVIDIA lifted the embargo on the GeForce RTX 3080 Founders Edition.

For unknown reasons, NVIDIA has decided to split the embargo between reviews for the Founders Edition and reviews of partner cards. Today we can only read reviews of NVIDIA's own model.

The NVIDIA Founders Edition for the Ampere series is not a reference design, so to speak. This graphics card will be sold exclusively by NVIDIA and will feature special features such as its own board design and Molex Microfit (12-pin) power connectors.

4 1 1

Links to reviews of NVIDIA GeForce RTX 3080 FE

3DMGame

4Gamer

Adrenaline

arhn.eu

ASCII

AusGamers

Bit-Tech

Bitwit

Benchmark

blackwhite TV

Comptoir Hardware

Computer Database

Cowcotland

El Chapuza sInformatico

Eurogamer / DigitalFoundry

Expreview

Gamers nexus

Gaming Nexus

GDM

Geeknetics

Gear Seekers

G

Greg Salazar

Guru3D

Forbes

HardwareBattle

Hardware Canucks

HardwareLuxx German

HardwareLuxx English

Hardware Unboxed

Hardware Upgrade

Hot Hardware

Hexus

igor'sLAB

io-tech

IT Media

Jagat Review

JayzTwoCents

Joker Productions

Kitguru

Lab501

Linus Tech Tips

Lowyat

news.mynavi

3D overclock

Overclockers Club

Paul's Hardware

PC Games Hardware

PCMasterRace LA

PC Perspective

PC Watch

PurePC

SweClockers

TechGage

TechPowerUP

Tech Of Tomorrow

TechSpot

Techtesters

The FPS Review

The Tech Chap

Tech YES City

Tom's Hardware

Tweakers

TweakTown

Wccftech

xanxogaming

x-com

XFASTEST Hong Kong

XFASTEST Taiwan

ZOL

Add a comment

Photos of AMD Radeon RX 6000 video cards with 3 and 2 fans

YouTuber JayzTwoCents has posted photos and renders of a graphics card with AMD RDNA2 architecture. The photo and two renders confirm the design presented by AMD.

Two renders of the Radeon RX 6000 series by JayzTwoCents show two versions of the RX 6000 cooler design. The first render uses the same triple-fan cooler seen on Twitter, where AMD officially showed it off for the first time. Jay also provided an actual photo of the card. Later in the video, he also shows a dual-fan version of the same cooler, likely a cut-down Big Navi (21) version.

He suggests that the three-fan model could be launched as the Raden RX 6900 series, with the latter under a different series (6800, 6700). Interestingly, the dual-fan version also has two 8-pin power connectors.

AMD is set to announce the Radeon RX 6000 series on October 28th. AMD will most likely confirm the detailed specifications and release date of the RDNA2 series at the event.

 3 1

AMD Radeon RX 6000 graphics card, Source: JayzTwoCents

 3 2 3 3

AMD Radeon RX 6000 graphics card, Source: JayzTwoCents

 

Add a comment

AMD RDNA2 will get AV1 decoding

AMD RDNA2 will receive AV1 decoding, Sienna Cichlid IDs are provided

AMD has introduced a Linux kernel patch with AV1 codec registers. This should confirm that the future RDNA2 architecture will support AV1 decoding on Radeon series desktops (RX and Pro). Both Navi 21 (Sienna Cichlid) and Navi 22 (Navy Flounder) will have Video Core Next 3.0 implementation.

AV1 decoding support becomes standard. Intel Tiger Lake series with Xe-LP architecture already supports AV1 decoding. The just-announced NVIDIA GeForce RTX 30 series will also feature full AV1 acceleration via a built-in hardware decoder.

Microsoft is adding the AV1 Video Extension, which should improve AV1 performance in Windows 10 with higher resolution. Google also updated its Chrome web browser in Windows 10 to support AV1 hardware decoding.

AV1 is a new, open, free video encoding format with up to 50% higher bitrate than H264, designed to be the successor to Google's VP9 and a competitor to H265 (HEVC). AV1 should be 30% more efficient than VP9 and HEVC.

2 1 1 

AMD Linux patches adding AV1 registers

 

Sienna Cichlid device IDs listed

Big Navi, also known as Navi 21, also sees its first IDs added to the core. It's unclear if there are any differences between the Sienna Cichlid and the Navi 21 (aka Raden RX 6000). The first one was seen with variables related to HBm2 technology, but has been relatively quiet on the topic since then. The Radeon RX 6000 series is expected to fully utilize GDDR6 memory technology.

AMD is set to announce the Radeon RX 6000 series on October 28th.

2 2 

Sienna Cichlid Device IDs, Source: Freedesktop

Add a comment

Meet the Intel Core i7-10870H and Core i5-10200H

Intel is releasing two new Comet Lake-H processors.

There have been no official announcements, but Intel has added two SKUs to the Comet Lake-H lineup. A fourth 8-core high-performance processor has been launched according to Intel's own website.

Interestingly, the launch of the processor allegedly arose due to problems with the delivery of the older Core i7-10875H model. This processor is very popular among high-end gaming laptops. After all, this is the fastest Core i7 processor in the lineup.

Compared to the 10875H, the 10870H offers a 100 MHz lower base clock. This doesn't seem like a big difference, but it might be worth considering since many enthusiasts are often advised to disable Intel Turbo. This is because turbocharging does not provide a significant performance improvement, but it does so at the expense of much higher temperatures.

Today on Reddit, the Schenker XMG team announced upcoming supply issues for the higher-end processor model (10870H). It is worth noting that XMG laptops (Neo 15/17) should not experience supply problems over the next 2-3 months:

[…] There is a CPU supply issue on the horizon for the i7-10875H. This situation is similar to that on desktops, where the Intel Core i9-10900K is all out of stock locally and is being replaced in most sales channels by the new Intel Core i9-10850K, which has almost the same performance (same cores, threads, cache, etc.) etc.), but the clock frequency is 0,1 GHz lower. […]

In the meantime, it looks like we have enough CPUs that we can continue to ship the i7-10875H to XMG NEO over the next 2-3 months. I’m not sure yet what will happen next, we’ll see.

XMG Schenker Announcement on Reddit

 

In addition, Intel has also launched a new quad-core Core i5-10200H processor, which is the slowest processor in the mobile H series. This processor has a 100 MHz lower base frequency than the 10300H and a 400 MHz lower turbo frequency.

 1 1

Intel comparison page for all Core 109xx and 108xx processors:

 1 2

Intel 10th Generation Comet Lake-H High-Performance Processor Series

 1 3

1 4

Intel Core i5-10200H and Core i7-10870H

Add a comment

NVIDIA GeForce RTX 3080 performance on Ethereum

Cryptocurrency performance of NVIDIA GeForce RTX 3080

Test results for GeForce RTX 3080 in NiceHash mining have been provided. RTX 3080 tested at 73-74 MH/s with DaggerHashimoto algorithm, depending on the miner.

On the output screen the card reaches about 81 MH/s. With maximum memory overclocking, the hashing speed increases to 84-89 MH/s. This is at 100% maximum power. Values ​​for other power limits are included below.

In comparison, the Radeon RX 5700 series offers around 50-54 MH/s in ETH mining DaggerHashimoto, which is also comparable to the GeForce RTX 2080 Ti at around 54 MH/s.

Dagger-Hashimoto is the initial version of the Ethreum Ethash mining algorithm. This is a memory-intensive algorithm designed to be ASIC-resistant.

 5 1

5 2

Note that this was just a quick attempt to get the hash values, not a full long-term test:

 5 3

NVIDIA GeForce RTX 3080 Stock

A hashrate of 92-93 MH/s was only observed for a short period of time.

 5 4

Memory NVIDIA GeForce RTX 3080 OC

Add a comment

NVIDIA GeForce RTX 3090 and RTX 3080 reference boards

Both NVIDIA Ampere reference boards have been pictured.

NVIDIA GeForce RTX 3080 reference PCB

NVIDIA is introducing two graphics card models based on the PG132 reference board: the RTX 3090 and RTX 3080. Both cards will also be available with NVIDIA Founders Edition-only boards (PG133 for the RTX 3080 FE and PG136 for the RTX 3090 FE).

Both reference boards have been photographed and we can take a closer look at the differences between the boards.

 4 1

The main difference between the RTX 3080 and RTX 3090 board is the lack of NVLink multi-GPU header. Only the RTX 3090 supports dual GPU configurations.

Another distinctive feature is the absence of two GDDR3080X front memory modules on the RTX 6 board. The RTX 3090 has dual-sided memory, which means a total of 24 8 Gbit (1 GB) modules. The RTX 3080, on the other hand, has just 10 modules. A 3080GB variant of the RTX 20 is also planned, but based on information from AIB, this variant will debut after the Big Navi.

Both cards have dual 8-pin power connectors by default, but there are some changes to the components. It looks like the RTX 3080 has lost two power stages compared to the RTX 3090. We also counted 6 capacitors missing from the RTX 3090.

As you may have noticed, even the RTX 3090 board does not have all power stages and capacitors installed. This board can also be used for other cards in the future such as TITAN Ampere.

 4 2

NVIDIA PG132 reference board for RTX 3080, Source: Bykski

NVIDIA GeForce RTX 3090 reference PCB

 4 3

NVIDIA PG132 reference board for RTX 3090, source: Yuuki_Ans

NVIDIA GeForce RTX 3080 'Founders Edition' PCB

 4 4

NVIDIA Founders Edition PG133 Board for RTX 3080

Add a comment

Intel DG2 discrete graphics with 6GB and 8GB GDDR6 memory

The Intel DG2 was listed with 8GB of GDDR6 memory in the specification sheet for an unspecified device. The product is possibly a high-end mobile device, probably a gaming laptop, which will hit the market next year.

The product will be paired with the yet to be announced Tiger Lake-H series. This series will be the first series of high-end mobile devices from Intel (45W) with Willow Cove core design.

Optionally, the product will also be offered with Intel Alder Lake-P. According to our information, this product will be released in late 2021 and will also feature the Alder Lake-H and Alder Lake-U series. However, unlike the Tiger Lake-H series, Alder Lake will feature a new hybrid architecture with large and small cores. The first wave should be expected with 2 small and 8 large cores. The 6+8 variant will be released later, sources told VideoCardz.

The 2GB DG8 variant is possibly a 512 EU variant, also featuring a 256-bit memory interface.

 3 1

Upcoming product with Tiger Lake-H, Alder Lake-P and Intel DG2 graphics

 

Intel DG2 384EU GPU, 6GB GDDR6?

The PCB diagram of the DG2 384EU variant was sent to us recently. Its size is about 189 mm2. The PCB looks densely packed with components and could in fact be a discrete MXM variant for laptops.

This DG2 model has 6 GDDR6 memory modules, indicating a 192-bit memory interface.

 3 2

Schematic diagram of Intel DG2 384 EU

 

What's also interesting is that the DG2 may also have a USB Type-C connector. It's important to note that the DG2 series, as mentioned above, is designed for laptops. The most recent Intel laptops now have a Thunderbolt 4 connector, which can also act as a display connector, so the entry may refer to this connector.

Yesterday, AMD announced that its high-end Big Navi design will also have a USB Type-C connector, while NVIDIA has not adopted USB Type-C (Virtual Link) for its GeForce RTX 30 series.

3 3

Add a comment

Intel Xe-HPG graphics cards offer up to 960 EU execution units

A series of new generation discrete video cards from Intel.

2 1

Xe-HPG architecture advertised by Intel

 

According to rumors received from a Twitter user, Intel is allegedly evaluating the possibility of launching a 960 Execution Unit chip based on the Xe-HPG architecture.

The purported Intel DG2 discrete graphics could be designed not only for desktop gaming, but also (or primarily) for the mobile discrete platform. There is also a possibility that there will be two lines based on the DG2 series. In other words, there are still more questions than answers.

The DG2 series is currently planned to offer 128, 384 and 512 execution modules. It's unclear if each of these configurations comes with a separate die, but the 384 is rumored to be between 185mm and 188mm in size2, which is a fairly small processor.

According to the source, Intel is also evaluating a 960 EU variant, but it is unclear whether this GPU will go into mass production. This configuration will definitely put pressure on high-end NVIDIA Ampere and AMD RDNA2 graphics. The Xe-HPG GPU with 960 EU will offer 7680 FP32 ALUs (cores).

Like AMD's RDNA2, Intel's Xe-HPG lineup will focus on GDDR6 memory technology, leaving GDDR6X exclusive to NVIDIA Ampere. The 384 EU die is rumored to offer a 192-bit bus and 6GB of GDDR6 memory.

Intel Xe-HPG is currently scheduled for Q2021 XNUMX. Intel has confirmed that Xe-HPG will be produced at an external foundry, the manufacturing site has not yet been confirmed.

 2 2

Intel Xe-HPG will appear in the second quarter of 2021.

 

Xe-HPG has one SKU with 960EU, but it is not known whether this will be mass produced.

384EU DG2 die area
≈185-188 mm2 ~192bit 6GB GDDR6~

960 EU Xe-HPG expected to be evaluated

 2 3

User MebiuW comments on a possible DG2 manufacturing unit

Add a comment

AMD Shows Off Radeon RX 6000 Graphics Card Design

AMD has officially confirmed the design of its Radeon RX 6000 graphics card.

AMD Radeon RX 6000: is it Big Navi RX 6900XT?

1 1

AMD Radeon RX 6000 Graphics Card (based on Navi 2X - RDNA2 GPU)

 

AMD showed off its Radeon RX 6000 graphics card. This model will likely use a Big Navi GPU with RDNA2 architecture.

The video card has a dual-slot design with three axial fans. The card also has up to two 8-pin power connectors. Such configurations provide power up to 375 W. AMD has not confirmed the names of its Navi 2X SKUs, but we can assume that it is most likely the Radeon RX 6900X. This is also the first Radeon graphics card to feature the new Radeon logo.

The graphics card has four display connectors: 2x DisplayPort (probably 1.4 with DSC), 1x HDMI (probably 2.1) and USB Type-C.

1 2

AMD has provided virtual coordinates to find the graphics card on Fortnite's AMDCreative Island: 8651-9841-1639. If you find anything, be sure to post it in the comments. Some screenshots have already been provided by @NateOrb:

1 3

1 4

More screenshots from @boot_sequence:

1 5

Video from Anshel Sag:

Add a comment

NVIDIA GeForce RTX 3080 Founders Edition: Thermal Tests

One reviewer accidentally uploaded footage from FLIR (thermal) testing of the GeForce RTX 3080 Founders Edition. The video was quickly deleted.

The video showed the heating of the NVIDIA GeForce RTX 3080 Founders Edition graphics card in an 8-minute video (8x acceleration).

The heatsink reached 65 degrees Celsius compared to 34,4 degrees Celsius in standby mode. This is the temperature of the heatsink that the cursor is pointing at, not the maximum temperature of the card. The PCB is clearly hotter, but the values ​​are not specified.

This test doesn't really say anything because the original video has been removed and it could have had a description. It is unclear what the exact testing procedure was and what software was used for this test.

It's also worth noting that the fan on the RTX 3080 Founders Edition doesn't spin. This model uses 0dB technology, where the card will not use fans at a certain temperature level.

 4 1

Temperature test of NVIDIA GeForce RTX 3080 (idle mode - left, load - right)

Add a comment