Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]

Status
Not open for further replies.
Was there an actual bench referenced or is that "rough" in the sense of using the nearest PC GPU?
 
Was there an actual bench referenced or is that "rough" in the sense of using the nearest PC GPU?
My thoughts exactly.

Sounds like the PS4's rough score of 5000 is simply how much the 1.76 TFLOPs HD7850 with 155GB/s bandwidth and 32 ROPs (very similar throughput to Liverpool) scores on the 3dmark 11 Performance GPU test:

https://www.legitreviews.com/xfx-radeon-hd-7850-1gb-core-edition-video-card-review_2048/9

YOHreHy.jpg





Tum Apisak is a reliable data miner to find new CPU/APU codenames and interpreting them, but I don't know of him actually having insider sources that would tell him they have PS4 systems running Windows.
 
Tum Apisak is a reliable data miner to find new CPU/APU codenames and interpreting them, but I don't know of him actually having insider sources that would tell him they have PS4 systems running Windows.
Are there alternative ways of submitting results that we are unaware of ? Or support for varying OS/benchmarking for companies etc at the professional level?
 
At the same time, would there be any reason to think a PS4 running the benchmark would score radically different?
 
Are there alternative ways of submitting results that we are unaware of ? Or support for varying OS/benchmarking for companies etc at the professional level?

I can't find anything on their website about this, but maybe UL Benchmarks takes custom orders to get their software running on devices with fully custom OSes and hardware like the PS4. Though if I were Sony I'd probably be much more interested in benchmarks running their own ICE engines or popular 3rd party engines like the Unreal Engine (e.g. the Samaritan demo).
In practice, 3dmark only serves to compare GPUs in the PC space. Console makers are probably more interested in figuring out how much eye candy they can put at a given resolution and framerate.


At the same time, would there be any reason to think a PS4 running the benchmark would score radically different?
Assuming it's the graphics score which doesn't depend a lot on the CPU then no, it'd probably score really close to a HD7850.

But assuming the PS4 would score close to 5000 in that test isn't what's fishy here.
What's fishy is claiming this Gonzalo APU is a custom SoC for the PS5 that somehow ran Windows with functional drivers in a popular benchmark for PC, and somehow they did the same with a PS4.

If this was someone claiming the PC-based devkit with off-the-shelf hardware that approaches the performance of the final console was getting that score, I'd find it a lot more believable.
 
I wrote a script to find Firestrike scores with unknown CPU and GPU from April 9th to April 11th. These are my results after 4 hours of running. I stopped the script because I think I made my point. They are all AMD Gonzalo!
https://www.3dmark.com/fs/19010366
https://www.3dmark.com/fs/19010375
https://www.3dmark.com/fs/19010540
https://www.3dmark.com/fs/19010615
https://www.3dmark.com/fs/19011253
https://www.3dmark.com/fs/19011424
https://www.3dmark.com/fs/19011521
https://www.3dmark.com/fs/19011546
https://www.3dmark.com/fs/19011942
https://www.3dmark.com/fs/19011979
https://www.3dmark.com/fs/19012048
https://www.3dmark.com/fs/19012170
https://www.3dmark.com/fs/19012298
https://www.3dmark.com/fs/19012509
https://www.3dmark.com/fs/19012564
https://www.3dmark.com/fs/19012600
https://www.3dmark.com/fs/19012837
https://www.3dmark.com/fs/19013194
https://www.3dmark.com/fs/19013294
https://www.3dmark.com/fs/19013397
https://www.3dmark.com/fs/19013734
https://www.3dmark.com/fs/19013756
https://www.3dmark.com/fs/19013789
https://www.3dmark.com/fs/19013819
https://www.3dmark.com/fs/19013946
https://www.3dmark.com/fs/19013972
https://www.3dmark.com/fs/19014428
https://www.3dmark.com/fs/19014470
https://www.3dmark.com/fs/19014639
https://www.3dmark.com/fs/19014681
https://www.3dmark.com/fs/19014737
https://www.3dmark.com/fs/19014753
https://www.3dmark.com/fs/19014809
https://www.3dmark.com/fs/19014823
https://www.3dmark.com/fs/19014861
https://www.3dmark.com/fs/19014871
https://www.3dmark.com/fs/19014882
https://www.3dmark.com/fs/19014949
https://www.3dmark.com/fs/19014966
https://www.3dmark.com/fs/19015004
https://www.3dmark.com/fs/19015112
https://www.3dmark.com/fs/19015217
https://www.3dmark.com/fs/19015346
https://www.3dmark.com/fs/19015571
https://www.3dmark.com/fs/19015957
https://www.3dmark.com/fs/19015967
https://www.3dmark.com/fs/19016060
https://www.3dmark.com/fs/19016149
https://www.3dmark.com/fs/19016200
https://www.3dmark.com/fs/19016228
https://www.3dmark.com/fs/19016286
https://www.3dmark.com/fs/19016364
https://www.3dmark.com/fs/19016374
https://www.3dmark.com/fs/19016419
https://www.3dmark.com/fs/19016477
https://www.3dmark.com/fs/19016550
https://www.3dmark.com/fs/19016646
https://www.3dmark.com/fs/19016955
https://www.3dmark.com/fs/19017007
https://www.3dmark.com/fs/19017028
https://www.3dmark.com/fs/19017061
https://www.3dmark.com/fs/19017080
https://www.3dmark.com/fs/19017255
https://www.3dmark.com/fs/19017332
https://www.3dmark.com/fs/19017371
https://www.3dmark.com/fs/19017409
https://www.3dmark.com/fs/19017461
https://www.3dmark.com/fs/19017561
https://www.3dmark.com/fs/19017628
https://www.3dmark.com/fs/19017921
https://www.3dmark.com/fs/19018013
https://www.3dmark.com/fs/19018155
https://www.3dmark.com/fs/19018183
https://www.3dmark.com/fs/19018186
https://www.3dmark.com/fs/19018252
https://www.3dmark.com/fs/19018590
https://www.3dmark.com/fs/19018680
https://www.3dmark.com/fs/19018830
https://www.3dmark.com/fs/19018874
https://www.3dmark.com/fs/19019001
https://www.3dmark.com/fs/19019186
https://www.3dmark.com/fs/19019256
https://www.3dmark.com/fs/19019290
https://www.3dmark.com/fs/19019360
https://www.3dmark.com/fs/19019461
https://www.3dmark.com/fs/19019538
https://www.3dmark.com/fs/19019611
 
The list that I scraped are just garbage results that any database would have. Maybe 0.1% of the results on 3dmark is garbage. Someone found one by chance that had a score over 20k.

I randomly clicked one of your links and the score was 28.3k. I know we can do better.
 
Last edited:
Node.js code if anyone is interested in running the script in a node environment. You'll need to setup up the dependencies in package.json file.

Code:
const { parse } = require('node-html-parser');
const axios = require('axios');

const MIN_ID = 19019611;
const MAX_ID = 19025000;
const URL = "https://www.3dmark.com/fs/";
const SLEEP_MS = 250;
const UNKNOWN = 'Unknown';

const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

const scrape = async () => {
  for (let id = MIN_ID; id < MAX_ID; id += 1) {
    const url = `${URL}${id}`;
    try {
      const { data } = await axios.get(url);
      const root = parse(data);
      const resultHeader = root.querySelector('span.result-header-small');
      const gpu = resultHeader.childNodes[3].text.trim();
      const cpu = resultHeader.childNodes[5].text.trim();

      if (gpu.indexOf(UNKNOWN) !== -1 && cpu.indexOf(UNKNOWN) !== -1) {
        console.debug(`Found: ${url}`);
      }
      await sleep(SLEEP_MS);
    } catch (err) {
      console.error(`Error in ${url}: `, err);
    }
  }
};

(async () => {
  try {
    await scrape();
  } catch (e) {
    console.error(e);
  }
})();
 
Status
Not open for further replies.
Back
Top