DirectCompute Benchmark for Vista SP2 with DX11 or Win7


Recommended Posts

Ok, it appears he did change how the application works.

Apparently (from what I can tell) if you DO NOT have OpenCL support, his program refuses to give DirectCompute results. Upon installing the Stream 2.0 SDK from AMDs site, it is now generating DirectCompute results. IMO that is the wrong way to do it, if a GPU supports DirectCompute (like the 4000 series with 9.12 drivers) then generate a DirectCompute Score. If it supports OpenCL (such as the Stream 2.0 SDK from AMD) then generate OpenCL Scores. Do not limit the program by ignoring DirectCompute scores because it doesn't support OpenCL (they are two different things).

i7-920, overclocked to 4Ghz.

ATI 4890 Toxic, overclocked to Atomic speeds (slight bump in core frequency).

post-168771-1262476206_thumb.jpg

I doubt it, at least not exclusively.

DirectCompute Benchmark 0.35: Generates a DirectCompute Score without a problem.

DirectCompute Benchmark 0.45b: Refuses to generate a DirectCompute Score.

How exactly is that "On my end"?

Because it works for others with .45 and ati...

edit: nvm

Results for version 0.45b:

NVIDIA GeForce 9800 GT @ 1753 MHz (10DE / 614 / 23721682)

Intel® Core2 Quad CPU Q6600 @ 2.40GHz (4 logical CPUs)

Benchmark version: v0.45

DirectCompute: D864.7

OpenCL: C603.2

CPU: M39.7

NVIDIA Windows Kernel Mode Driver, Version 195.62

nvlddmkm 8.17.11.9562

Windows 7 x64 Ultimate Edition (build 7600)

Doesn't look that bad for a two year old system. (the 9800 GT is really a 8800 GT with the BIOS modded)

Because it works for others with .45 and ati...

edit: nvm

I know you have already seen my post on how I fixed the issue, but that statement is too general IMO. Just because a program works for others, doesn't mean its going to work normally for everyone. Thats like saying no program in the world contains bugs, any issue someone has is due to their own mistake.

I know you have already seen my post on how I fixed the issue, but that statement is too general IMO. Just because a program works for others, doesn't mean its going to work normally for everyone. Thats like saying no program in the world contains bugs, any issue someone has is due to their own mistake.

Well I was right, your system didn't have something installed causing it to not work :shiftyninja:

Your answer makes sense, I already had the 9.12 hotfix with the stream SDK installed before i benchmarked 0.45 so I didn't notice that.

Odd thing is, I too installed the 9.12 hotfix (I checked my driver version with the 9.12 hotfix driver version and its an exact match), but I did not have the Stream SDK installed. Its possible I selected to not install it though, if it was possible. When I install the drivers I prefer to go with the bare minimum (I do not install the HDMI audio, as I do not use HDMI, etc).

ViperAFK, we were both right. The author of DirectCompute Benchmark DID change something which prevents the program from using DirectCompute if you are missing OpenCL, that was a "feature" not there in previous versions of DCB. At the same time, you were right in saying it was something with my setup.

What I think is more interesting is WDDM 2.0 and whether it will be coming in Windows 8 or Windows 9 - that will be when directcompute and other features start showing some major leap forwards in terms of the GPU being put to some major use. I hope we'll start seeing some vendors utilise it to speed up video compression speed.

Odd thing is, I too installed the 9.12 hotfix (I checked my driver version with the 9.12 hotfix driver version and its an exact match), but I did not have the Stream SDK installed. Its possible I selected to not install it though, if it was possible. When I install the drivers I prefer to go with the bare minimum (I do not install the HDMI audio, as I do not use HDMI, etc).

ViperAFK, we were both right. The author of DirectCompute Benchmark DID change something which prevents the program from using DirectCompute if you are missing OpenCL, that was a "feature" not there in previous versions of DCB. At the same time, you were right in saying it was something with my setup.

The Stream SDK isn't included with the drivers, you have to install it seperately to get the OpenCL component (stupid I know) but I would imagine that the OpenCL component will probably be included in future driver releases

What I think is more interesting is WDDM 2.0 and whether it will be coming in Windows 8 or Windows 9 - that will be when directcompute and other features start showing some major leap forwards in terms of the GPU being put to some major use. I hope we'll start seeing some vendors utilise it to speed up video compression speed.

Well people have already written lots of CUDA enabled encoders, they'll eventually move to OpenCL (yay for multi-platform) or DirectCompute (yay for Windows only!)

Well people have already written lots of CUDA enabled encoders, they'll eventually move to OpenCL (yay for multi-platform) or DirectCompute (yay for Windows only!)

Sorry, I'm a bit out of the loop; is there a h264/mp4/aac encoder that takes advantage of directcompute?

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • I was expecting the end of the world to happen before this game or elder scroll 6 to come out.
    • OpenAI and Broadcom unveil Jalapeño, a new AI chip built for LLM inference by Pradeep Viswanathan Image by OpenAI Thanks to the exponential growth of ChatGPT and other LLM-based applications, NVIDIA has grown from a $200 billion company into the first public company to reach a $5 trillion market cap. Even though hyperscalers such as Google and Amazon have their own mature AI accelerators, NVIDIA still dominates the AI infrastructure market with multiple generations of GPUs. Microsoft, OpenAI, and Meta remain among NVIDIA’s largest customers, while Google and Amazon continue to be significant NVIDIA customers as they serve AI workloads for customers on their cloud platforms. Today, OpenAI and Broadcom announced Jalapeño, OpenAI’s first custom “Intelligence Processor” designed specifically for large language model inference. The new chip is the first product from a multi-generation compute platform being developed by OpenAI. OpenAI highlighted that Jalapeño was built from the ground up for current and future LLM workloads, rather than being a general-purpose accelerator adapted for AI. Despite heavy competition from Gemini, Claude, Copilot, and others, ChatGPT remains the most used AI platform in the world. OpenAI mentioned that it leveraged its knowledge of how its models and products run at scale, including ChatGPT, Codex, the API, and future agentic AI systems, to design this new chipset. Its chip architecture reduces data movement while balancing compute, memory, and networking resources. Jalapeño will be deployed in production systems starting in late 2026; however, engineering samples are already running machine learning workloads in OpenAI’s labs at production target frequency and power. According to its internal testing, OpenAI claims this chip can deliver “substantially better” performance per watt, and a detailed technical report is expected in the coming months. While OpenAI designed the chip, Broadcom handled silicon implementation and networking technologies, including Tomahawk networking silicon, and Celestica is assisting with board, rack, and system-level integration. OpenAI pointed out that Jalapeño went from initial design to manufacturing tape-out in just nine months, which it claims is the fastest ASIC development cycle achieved for a high-performance advanced semiconductor. The company attributed the speed of development to its own LLMs, which were used during the chip design and optimization process. Broadcom CEO Hock Tan stated that the company's plan is to deploy the Jalapeño platform at a gigawatt scale with Microsoft and other partners starting in 2026. With Jalapeño, OpenAI joins Google, Microsoft, and Amazon to become a full-stack AI player. The company already develops models and products, and is now moving deeper into infrastructure, including chips, kernels, networking, scheduling, and deployment systems.
    • I'm aware. That information should have been included in the article, making it more complete and information.
    • Converseen 0.15.2.5-2 by Razvan Serea Converseen is a free and open-source batch image converter and resizer. It supports over 100 formats, including DPX, EXR, GIF, JPEG, JPEG-2000, PNG, SVG, TIFF, WebP, HEIC/HEIF, and many others. Users can convert, resize, rotate, flip, and compress multiple images at once. It can also transform entire PDF documents into individual image files. Powered by the ImageMagick library, Converseen features a user-friendly interface and is available in both installer and portable versions. Here’s a list of all the features you can find in Converseen: Batch image conversion (supports 100+ formats) Resize images in bulk Rotate and flip images in bulk Compress images to reduce file size Convert entire PDF documents into image files Support for multiple image formats (JPEG, PNG, TIFF, PDF, BMP, GIF, and more) Customizable output settings (quality, resolution, etc.) Image effects and adjustments (such as brightness, contrast, etc.) Convert images to PDF User-friendly graphical interface Support for drag-and-drop functionality Extract an image from a Windows icon file (*ico) Supports adding watermark to images Portable and installer versions available Leverages ImageMagick for processing power Allows renaming of images in bulk Supports EXIF data editing (for JPEG images) Easy-to-use GUI for non-technical users Command-line support for advanced users Free and open-source software Cross-platform availability Available in multiple languages Download: Converseen 0.15.2.5-2 | Portable | 32-bit | ~40.0 MB (Open Source) View: Converseen Homepage | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • Regarding the AI photo, I LOVE AI in that regard, you ask it what you want and it gives you a lovey photo in under a minute, that would taken me an hour to make in photoshop and it wouldn't have looked nearly as good. 2 nights ago I spent a couple hours collaborating with AI.  I did not say write me an article. I would write one or 2  paragraphs, then I would ask it to clean it up so it read better but still keeps the information I was trying to convey.  Rinse repeat.  
  • Recent Achievements

    • First Post
      Tom Schmidt earned a badge
      First Post
    • One Month Later
      D0nn13 earned a badge
      One Month Later
    • Rookie
      +ChiefOfNeo went up a rank
      Rookie
    • One Year In
      Tom Schmidt earned a badge
      One Year In
    • One Month Later
      Tom Schmidt earned a badge
      One Month Later
  • Popular Contributors

    1. 1
      +primortal
      448
    2. 2
      +Edouard
      176
    3. 3
      PsYcHoKiLLa
      123
    4. 4
      Michael Scrip
      81
    5. 5
      Xenon
      75
  • Tell a friend

    Love Neowin? Tell a friend!