Jump to content
Search In
  • More options...
Find results that contain...
Find results in...


  • Content Count

  • Joined

  • Last visited


This user doesn't have any awards

1 Follower

About randomhkkid

  • Title

Contact Methods

  • Twitch.tv

Profile Information

  • Gender
    Not Telling
  • Location
    United Kingdom
  • Interests
    Teaching machines to learn
  • Biography
    Engineer | Photographer
  • Occupation


  • CPU
    i5 4690K 5ghz
  • Motherboard
    MSI Z97M Gaming
  • RAM
    16GB Corsair Dominator Pro 2400mhz
  • GPU
    GTX 980 1.5ghz
  • Case
    350D with Window and RGB lighting
  • Storage
    RAID0 2x 256GB Crucial MX100, 128GB Sandisk Extreme, 2x 1TB Seagate SSHD
  • PSU
    Cooler Master V850
  • Display(s)
    4K 60Hz 27" Acer CB271HK
  • Cooling
    Corsair H105 Push-Pull
  • Keyboard
    Thermaltake Poseidon Z RGB Blue switch
  • Mouse
  • Operating System
    Windows 10 Creators

Recent Profile Visitors

1,494 profile views
  1. Ah my bad, was under the impression that INT32 + FP32 was the async compute feature. Amended.
  2. Yeah that's why I added the second chart in the first post. According to Nvidia without DLSS there's about a 20% loss of performance.
  3. Starting with a driver in April DXR capabilities will be enabled on non-RTX cards from the current and previous generation. Non-RTX Turing amd Volta will utilise INT32 + FP32. Pascal series cards will use FP32 only. Nvidia gives us an indication of the performance we can expect rendering a single frame of Metro Exodus. They compare roughly equivalent cards in the RTX 2080 and the GTX 1080Ti. A rough visual estimate gives: * INT 32 + FP32 - 2.5x slower * FP32 - 3.2x slower Converting this to frametimes and framerate assuming the bottom one is 16.66ms 60fps gives: * INT 32 + FP32 - 41.65ms, 24fps (cinematic af) * FP32 - 53.31ms, 19fps A second visualisation shows that DLSS provides a roughly 20% boost to the RTX card's performance. Without it the 1080ti is still around 2.8x slower at the same settings. Hopefully this does pave the way for more adoption of DXR in the industry. With AMD incoming with DXR support there should be a good number of users with access to the technology. Unity and Unreal have announced support for DXR in upcoming releases of their engines. Beyond games this is also great news for a range of 3D rendering applications.
  4. Scores seem about right to me. My 4690k at 4.6Ghz and 2400mhz memory scores 1719 and 436. Linear scaling puts your score at 1271 and 322.
  5. Likely due to the avx workload being more strenuous than a non-avx benchmark/stress test.
  6. Multi: 2811 Single: 422 Razer Blade 15 Advanced 2018 i7 8750H
  7. 2811 / 422 Razer Blade 15 Advanced 2018 i7 8750H Interestingly the score relative to the 8700k stock has dropped compared to CB r15
  8. The servers behind Jibo's functionality are apparently being shutdown. Jibo owners are being notified of this by way of an audio message and a dance to some music. Interestingly going to Jibo's website shows a bunch of missing video links and no real info as to what the 'limited' functionality will be once the servers go offline. Edit: So the limited functionality appears to only be touch interaction for its local applications. No speach recognition beyond wake word detection.
  9. The common concern amongst people who know what they're talking about is that the additional heat from the LCD being on could negatively impact thermals. The implementation for overclocking actually requires the official charger to be plugged in due to increased power draw.
  10. Keep in mind that in the new cards raytracing is offloaded to the tensorcores. The 1080ti may be able to do 1 gigaray but that's with all of its cuda cores bruteforcing it.
  11. Glad to see that they have been able to bring performance in line with other similar laptops (XPS 15) in terms of sustained performance. Unfortunately I think a lot of manufacterers were caught unguarded to the delays of 10nm meaning that thermals are suffering across the board.
  12. Source: https://www.anandtech.com/show/12789/razer-launches-the-razer-blade-156 Razer Blade 15.6 Models GTX 1060 GTX 1070 CPU Intel Core i7-8750H 6 Cores, 12 Threads 2.2 GHz - 4.1 GHz, 45W TDP RAM 16 GB - 32 GB DDR4-2667 GPU NVIDIA GTX 1060 Max-Q 1280 CUDA Cores 6 GB GDDR5 192-bit NVIDIA GTX 1070 Max-Q 2048 CUDA Cores 8 GB GDDR5 256-bit Storage 512 GB - 2 TB PCIe NVMe SSDs Display 1920x1080 60Hz IPS matte 1920x1080 144Hz IPS matte (optional) 3840x2160 60Hz IPS with touch factory calibrated displays Wireless Intel Wireless-AC 9260 802.11ac 160MHz channel support MU-MIMO 2x2:2 Bluetooth 5 Connectivity Thunderbolt 3 USB-C x 1 USB 3.0 x 3 mDP 1.4 x 1 HDMI 2.0b x 1 Battery 80 Wh 200 W Adapter 80 Wh 230 W Adapter Dimensions 355 x 325 x 16.8 mm 14 x 9.3 x 0.66 inches 355 x 235 x 17.3 mm 14 x 9.3 x 0.68 inches Weight 2.07 kg / 4.56 lbs (non-touch) 2.15 kg / 4.73 lbs (touch UHD) Price $1899.99 and up $2399.99 and up Ouch that price though. Guess it's to be expected, glad to see another manufacturer committing to colour calibration from the factory. The lack of G-sync and HDR support is a little disappointing to see on a high end laptop, they seem to claim not to support the former due to wanting to extend battery life. Glad they went for the slim bezels without silly webcam placement. Notably the 1070 max-q version is a little thicker.
  13. Source: https://www.anandtech.com/show/12749/first-10nm-cannon-lake-laptop-spotted-online-lenovo-ideapad-330-for-449 Laptop available for $449 on a Chinese retailer JD https://item.jd.com/26395831446.html Intel Core i3-8121U10nm Cannon LakeDual CoreFour Threads2.2 GHz Base3.1 GHz Turbo (?)No Integrated Graphics I didn't realise intel were still shipping some 10nm parts. I guess they do have sufficient yield to ship some lower volumes of parts. Will be interesting to see the IPC and efficiency gains of the 10nm process. AFAIK these will be the same gains we see in the mainstream and high end parts when they do start to ship.
  14. Also computer scientists. Major headache trying to acquire card for Deep Learning at the moment. Trying to do a thesis here
  15. Source code just released for the pixel core in the new pixel devices. Hopefully we'll get an idea of performance in ML type task soon. https://android.googlesource.com/kernel/arm64/+/android-easel-intel-linux-4.7-oreo-mr1/drivers/misc/paintbox/paintbox.c#196