IanCutress

Almost OT, but the weekend after computex is the Canadian GP. I think it's on at 4am or something silly, but I want to watch it if it's on anywhere.

I assume the Galaxy OC event going to be the same as last year? Just HKEPC + a few others at their booth?

Good words there Steve. I would also add it depends on memory settings (I know for a fact one reviewer does not enable XMP, as they have used motherboards where XMP doesn't work, but lists in the test bed it's some ultra high end kit), what software you have installed (via Steam or Origin or Disk) and what motherboard software is also there. For reference I've been solely on Catalyst 12.3 (March 2012) and NVID 296.10 drivers for the GPU portion of mobo reviews since Ivy launch. I am upgrading now to 13.1 and 310.90, but I am going back through the last 4/5 Intel and AMD platforms in 1/2/3/4 GPU configurations to generate a f*ckton of back data for our readers. Currently 3 weeks in alongside doing other things, many CPUs tested, many platforms tested, over 500 data points. There's also statistical variation - I would hedge a bet that some reviewers only run the scene once. You really need to run a scene multiple times (at least 4) then take an average, or remove top and bottom scores then take an average. Then add statistical variance / standard deviation, if we were doing this with more professionalism. A histogram would be even better. Having the benchmarks with a timedemo is a godsend - it means I don't have to sit over my test bed 24/7 and can still continue doing other reviews. (With every manufacturer wanting 10x reviews done this week, maintaining high throughput but still covering 99% of the bases is important.) Personally I find FRAPsing a portion of the game, even if it is the same portion, to be frustrating and not representative of final performance. If you're recording a regular dull portion of the game repeatedly, that's not fair on readers - if it's a really active game, then you're screwed because you can't keep things consistent. Minimum frame rates can vary wildly depending on the scene as well, or if XYZ software in the background decides to update itself/probe network activity. Yes, it's true that some timedemos can also not be fully representative - but the user can run a public timedemo themselves if they want to see where they stand. There is one issue with timedemos I will touch upon. The Metro2033 timedemo, for example, runs better the first time you run it. In about 50% of the first runs, the results can be up to 3% better than normal. For this benchmark, I typically take the average FPS of four runs. Because of this issue, I do two sets of my four runs, and only take the second set of four (or third set if there's a large variation in results).

Time to fire up the SR2?

If anyone's dealing with header issues and cURL, here's how to do it: <?php // retrieve data via cURL $ch = curl_init(); curl_setopt($ch, CURLOPT_URL, "http://hwbot.org/api/benchmarks"); curl_setopt($ch, CURLOPT_HEADER, 0); curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1); $xml_content = curl_exec($ch); curl_close($ch); // now you can process the XML content (i.e. via a xml to array function) echo $xml_content; ?> Also @IMOG, here's some things I have done with another APIs: http://www.borandi.co.uk/BC2XS http://www.borandi.co.uk/wpe as well as a couple of things for academia

Set the update all system to process in batches of 100, with a 60 second wait command between.

Two additional suggestions Devroush: (a) can you add the output to include the current server time/date in each API call as well? This would help with locally caching some data to make local loading faster. It would also help with a simple API call to just return the server time/date. Then I can do a local if(current-cache > 24hr) {update();} routine. (b) another request for an API of all current hardware in the database, that can be queried by OEM, manufacturer, chipset, socket et al. Just to return the hardware strings for the searches. Or even better, just a link to the the output of the hardware search function when you put in '?q=' - as I mentioned, the old HWBot layout source code gave us access to this, but now it's hidden behind some wizardry Many thanks!

I've not got anything solid yet, haven't had the time. Actually building something that shows data like this, and coding it up, takes about half a week - mainly to do with organising queries, requests, and dealing with objects/arrays. But take this one for example (in pseudo code) 1) Query API for list of benchmarks 2) Loop through the benchmarks, query number 1 score for that benchmark (Y benchmarks of API calls) and store in array 2a) If CPU benchmark, iterate through number of cores, query number 1 score for X cores, store in array 2b) If GPU benchmark, iterate through number of GPUs, query number 1 score for X GPUs, store in array 2c) If memory benchmark, iterate through memory types (may require another API call) 2d) If SSD benchmark, etc. 3) Loop through the benchmarks again, query number 1 score for the country, store in new array 4) Display table of benchmarks, no.1 score vs. best country score (comparing and outputting arrays), and number of points each score gets. Also can display the kit used, CPU MHz, and everything else There is a 30 second limit on most PHP servers, so dealing with too many API calls in one script will cause the script to end - I find about 50 API calls to individual submissions before you hit the limit. I have plenty of ideas, just need time. Or money, that works too.

Doesn't matter either way for me at least, my line of code works wonders with what comes out by default.

It was a global invite to anyone that could attend, not to specific people! 99% sure I'll be in TW for a couple of weeks around that weekend, though I doubt I'll have anything to overclock given that my suitcase is usually full of suits and stuff for Computex itself. Sounds like I could wrap an interesting story for a news item / range of drunken interviews and call it a 'business expense'

Take the submission # at the start, take the submission # at the end Pick random number between the two If it qualifies, it wins, If not, pick new random number. Could be the 1st, could be the 500th shouldn't take too long

I've noticed a series of submissions like that. http://hwbot.org/submission/2335039 Belongs to 8-pack, but no points, doesn't seem to be linked to his account on best scores, doesn't seem to be linked to the team.

Hey dude, recalculate this. Worth some team scores

Another request, an hardware API at /api/hardware. I essentially want a list of all the CPU/GPU hardware options, similar to the search/?q= query used in the search box on the main site. I know there's a text file with all the hardware somewhere on the server that was referenced in the source of the front page, but now it's obfuscated behind a link redirect That text file would be enough

Another small thing, which occurs in the main search as well. If I parse a command for the best single GPU result for 3DMV, it includes all the dual GPU card results. This is because the numberOfVideocards and something relating to the database. Could you include a numberOfGPUs as well, which is numberOfVideocards * GPUs per card?

A few suggestions.. (a) Put in the help file that you can call something like http://hwbot.org/api/benchmarks to get all the benchmarks. Some benchmarkIds have no benchmark associated, and there's no easy xml list of them to iterate through. (b) Simple PHP code like '$benchmarks = simplexml_load_file('http://hwbot.org/api/benchmarks');' fails, because it's returning JSON Though when I put the address into an address bar, it's returning XML. Is there some form of redirect going on? The code above gives the following error, indicating JSON: Warning: simplexml_load_file(): {"errors":null,"results":[{"name":"3DMark2001 SE","id":"1","type":"GPU","country in C:\Program Files (x86)\EasyPHP-12.1\www\HWBot API - main benchmarks\index.php on line 7 I tried with an ?option=XML, still nothing. Going to use $benchmarks = json_decode(file_get_contents('http://hwbot.org/api/benchmarks')); instead, which parses the JSON into an object/array hybrid :thumb: © Could we also get to use the tag ?applicationId=1 on the benchmarks api?

Are those temperature numbers going to be able to cope with negative numbers or possible -#.INF coming from the sensors?

Will hopefully be going, not sure on spare time either side of the event though.

OK Data Mining is what I really wanted! (sorry on bailing on the Benchmark API testing, things got rough) Tried the link in the PDF, http://hwbot.org/api/submissions?regionCode=europe&league=enthusiast&application=3dmark11_- _performance&gpuCoolingId=3&pictureAttached=true&gpuFamily=geforce_600&gpuCoreFreq=1200&gpuCoreFreqOperator=less_or_equal&limit=1, got an error. Regarding "The table below clarifies each API. API requests can be made with regular HTTP GET or POST requests, with different GET paremeters / POST data. The response format can be XML or JSON, depending on the request header.", my requests will be done via PHP, meaning we need an ?option=XML tag in there as there isn't a request header. Will start looking into it a bit more after some work that needs doing But nicely done !

Awesome! 100K is great

Look on ebay and be ready to shell out the money on ES CPUs

Someone submits, the system does a recalc of the ranking, and then the user quickly deletes, the system doesn't automatically recalc the ranking. If no-one submits a new result or requests a recalc, it takes some time for it to come around in the casual daily recalcs.

Read the thread, interesting stance. My take: Pro OC needs a handful of benchmarks to focus on, so keeping the number of benchmarks with global points limited is a good idea. Through the year you can add X benchmarks with Global Points, but have a yearly vote that lasts 4 weeks on 'benchmark to drop global points' is a good idea, and take out X benchmarks from Global Point rankings. Keep all benchmarks that have had hardware points with HW Points, but maintain security. The OC League is a tricky one to deal with, being a combination of HW and GL points. You could do what we used to do, and limit HW points to 300 for the league - I know it's already limited to top 20 scores on the HW side, so maybe no change is needed. So if you add Cinebench (ST + MT), Catzilla, 3D12, that leaves three benchmarks to get rid of Global Points at the end of the year, voted by the community. Reggie has a good point regarding having a boundary which allows global points. Unless it's one demanded by all the sponsors, initially add in a benchmark, and when the popularity > 5000 (it that submissions in last 12 months?), give it global points automatically. Any benchmark that loses global points status won't be eligible again for globals for 24 months. I don't think there's a need for an overall system benchmark. Any that are created either take too long or are exploitable. They're also not that fun to watch, especially for an audience. Technically a good GPU benchmark still abusively stresses the system (CPU, GPU, PSU, Memories).

There's been numerous discussions about this, basically because certain users can set up a WC loop with ambients sub-zero. The general rule came down on the side of: - As long as the coolant in your loop is water, stick it under water cooling. - If you add anti-freeze to your cooler or something which freezes below 0, it comes under sub-zero. - You are allowed to use a TEC/phase change water chiller on a pure water loop and still keep it as water cooling. If ever the coolant touching the CPU goes below 0C, it becomes sub-zero. All-in-one liquid coolers use ethylene glycol, which technically is liquid below zero, so as long as they are unmodded it still comes under water cooling.

Insane. Congrats! Nice 2500K too. ASUS Direct CU II card on a Maximus V something?