AMD Presents New Horizon

I don't play DOTA and don't stream games - so can someone explain to me why a 6700K overclocked to 4.5GHz can't simultaneously play DOTA and stream it without dropping frames (badly) on the stream?

I used to livestream all the time, it's hard as hell on the machine depending on the codec and resolution you're streaming at, especially in demanding games (not that DOTA is demanding). For most BIG streamers, they do use 2 machines, 1 running the game, and running into a capture card of a 2nd thats doing the encoding and streaming.

EDIT: just as an edit, my W3690 (essentially a 990X) can only reliably stream at 720, and thats heavily game dependent.
 
I don't play DOTA and don't stream games - so can someone explain to me why a 6700K overclocked to 4.5GHz can't simultaneously play DOTA and stream it without dropping frames (badly) on the stream?
A friend of mine has a 6 core i7 and it doesn't keep up with games + streaming. Most streamers have 2 PCs. 1 for gaming and 1 for streaming.
 
Assuming the numbers look similar to today's demos when real reviews come out (and they OC decently), I may end up dumping my 5820k setup for Ryzen. Color me impressed so far.

Maybe this release will force Intel to start making hex/octacore chips in their mainstream lineup?
 
I'm liking what I'm seeing so far! I'm not quite ready to build a new PC but if Zen delivers on all fronts it may accelerate my plans on building a new one.
 
Assuming the numbers look similar to today's demos when real reviews come out (and they OC decently), I may end up dumping my 5820k setup for Ryzen. Color me impressed so far.
Dunno, my 5820K is running along fine at 4.4GHz still today. I guess if you needed the extra cores, sure, but not sure it would push me to change on a daily desktop unless I was encoding all day long.
 
Ok- I'm settled in and it looks like about 8 minutes to start.

Not exactly excited... but I've got nothing better to do at the moment.
 
I'm impressed and was very pessimistic before today. If AMDs top CPU ends up fitting somewhere in the Intel Broadwell-E range in price and performance, I'm sold. Intel releasing that 6950X @ $1,700 was just a big fuck you to consumers. Showed everyone what will happen if AMD goes under. Here's to the underdog.
 
I'm very excited for January! I might consider upgrading my i5 to a 6-core in the next year or two, if there are enough games using more than four cores, and the price is right :D
 
The insistence on demoing games at 4K frankly baffles and concerns me, other than that I am pleased to say I am pleasantly surprised by the handbrake result, I still don't feel like blender is worth a damn judging by the benchmark results on the blender website. Hopefully with a 3.4 GHz boost 4GHz at least will be easily doable, with a sensible price this could turn out to be great . They can easily undercut intel and sell the 8 cores at around $700 for the unlocked models or something.



3.4GHz base isn't too bad, I mean if they specified it's base it must at least have 400mhz headroom otherwise they're basically trolling; the AMD trolling platfrom

Demoing games at 1080p or 1440p would just have people wondering if 4K is out of reach for AMD or such, besides, New Horizon was a attempt to appeal to gamers, that means aside from the beginning demos, showing off the latest and greatest PC gaming can offer, i.e 4k, battlefront, BF1, (i cant remember the other game), streaming, and Nvidia cards to pointedly say, want to upgrade your CPU but not GPU? Thats possible, hint, hint.
 
So it definitely looks like a promising chip architecture. First gen might not be worth getting but I can see the ZEN+ as being the chip to get in a year or two unless Intel really does something amazing at 10nm.
 
The insistence on demoing games at 4K frankly baffles and concerns me, other than that I am pleased to say I am pleasantly surprised by the handbrake result, I still don't feel like blender is worth a damn judging by the benchmark results on the blender website. Hopefully with a 3.4 GHz boost 4GHz at least will be easily doable, with a sensible price this could turn out to be great . They can easily undercut intel and sell the 8 cores at around $700 for the unlocked models or something.



3.4GHz base isn't too bad, I mean if they specified it's base it must at least have 400mhz headroom otherwise they're basically trolling; the AMD trolling platfrom

I'd be really curious to see how intel reacts with pricing if zen delivers, it's not like Intel will ever have any problems price matching AMD. Plus still curious to see die sizes on Zen
Well, I have to remain pessimistic while wanting to be optimistic in my heart. AMD marketing has not exactly been on the forefront of pushing a "real" narrative for a while now.
 
I thought Intel + AMD abandoned the clock speed race years ago, and shifted to a core count model. Is there a bare bottom count I should be aware of today?
 
I'm pleasantly surprised. Does it have to be the best? Hell no! It is betting on multi-threading, performance/TDP, and performance/$

This could help close the gap and help them gain back some market share; now that they made something useful (again), the successor to ryzen could make things interesting.

I like the throttled based on thermal dissipation thing, as I have a dual-loop watercooling system that uses a 55 gallon drum full of water to cool my radiators :)
 
The insistence on demoing games at 4K frankly baffles and concerns me, other than that I am pleased to say I am pleasantly surprised by the handbrake result, I still don't feel like blender is worth a damn judging by the benchmark results on the blender website. Hopefully with a 3.4 GHz boost 4GHz at least will be easily doable, with a sensible price this could turn out to be great . They can easily undercut intel and sell the 8 cores at around $700 for the unlocked models or something.



3.4GHz base isn't too bad, I mean if they specified it's base it must at least have 400mhz headroom otherwise they're basically trolling


I honestly don't think they should undercut too much. That in turn will start a price war they can't afford probably. I am sure they both know it lol. I think AMD will be like at 500-600 and intel will probably knock theirs down to 750 or so given AMD truly caught up alot in IPC and over delivered here from the look of it. Some people will pay higher just cuz its intel, I am sure AMD is happy with that.

I think there is a reason you heard nothing about pricing. They won't charge as much as intel, but I am sure if they are over delivering here, they wont exactly hand it over to you lol. I am expecting 500-600 dollar range minimum for the top end part if it is indeed performing like they showed. Handbrake was particularly surprising to me. That shit really chokes my PC when I ran the bench..
 
Does clocks matter really that much?? If they can get close to parity, its a win for everyone.

It matters because it leaves Intel a very valuable ace card, with Kaby lake doing 5GHz comfortably and most Haswell chips at 4.5 quit comfortably it's not going to win over anyone with recent HEDT Intel CPUs unless it clocks to 4.5 and you own broadwell lol

I honestly don't think they should undercut too much. That in turn will start a price war they can't afford probably. I am sure they both know it lol. I think AMD will be like at 500-600 and intel will probably knock theirs down to 750 or so given AMD truly caught up alot in IPC and over delivered here from the look of it. Some people will pay higher just cuz its intel, I am sure AMD is happy with that.

I think there is a reason you heard nothing about pricing. They won't charge as much as intel, but I am sure if they are over delivering here, they wont exactly hand it over to you lol. I am expecting 500-600 dollar range minimum for the top end part if it is indeed performing like they showed. Handbrake was particularly surprising to me. That shit really chokes my PC when I ran the bench..

$600 is a huge undercut man, 6900k is like 1000$
 
Disappointed with the stream, they didn't really show much.
Lots of talk and some potentially cherry picked "comparisons" / gpu limited @ 4k gaming.

Also near the end, reminded me of the "amd vs intel" streaming tests that tek syndicate did a while back.
Just because it "can" play + cpu stream at the same time doesn't mean you would want to. Especially with gpu accelerated streaming like shadowplay/relive now.

Still need to see 1080/1440 gaming benchmark comparisons with min/max/avg fps from a non biased reviewer.
 
I honestly don't think they should undercut too much. That in turn will start a price war they can't afford probably. I am sure they both know it lol. I think AMD will be like at 500-600 and intel will probably knock theirs down to 750 or so given AMD truly caught up alot in IPC and over delivered here from the look of it. Some people will pay higher just cuz its intel, I am sure AMD is happy with that.

I think there is a reason you heard nothing about pricing. They won't charge as much as intel, but I am sure if they are over delivering here, they wont exactly hand it over to you lol. I am expecting 500-600 dollar range minimum for the top end part if it is indeed performing like they showed. Handbrake was particularly surprising to me. That shit really chokes my PC when I ran the bench..

You know AMD is just chomping at the bit to offer $300+ CPU and $400+ GPU competition. If they deliver on Vega and Ryzen then I would put money in a sportsbook on them throwing product in those categories.
 
01.jpg

I tuned in late so I am now watching the first part. Noticed AMD might want to hire a new proofreader. "Platfrom." :D
 
I just ran the benchmark again with real time process priority and I managed 52.13s. hwmonitor says my CPU maxed out at 0.91W Shintai , so I guess I'm going to upgrade in a decade or so
 
Never mind, looks like total BS tweet. Ignore
CzlgRTFWEAIBcbJ.jpg


Well... Someone has some explaining to do.

My 5820K at 4350 did 52s, a 5960x at 4125mhz did 44s so how the hell is a stock 6900k at 29s ???
.eJwFwVEOgyAMANC7cAAKxVbmbQgSNFNKoGYfy-6-977mGZfZzKHa5wawnzPL2O1UGakWW0XqVVI_p81yQ1JN-bhL0wk-Ei8rEjmMjC9yBEgxMAaPC4e4sncMT3s3-TTbWzW_PwewIt0.Na_beLajqw6y5xPXS-vv1u7Ml_Y
 
Definitely impressing stuff shown today. The shift of the discussion from "will it match Sandy bridge i5" to "will it beat i7-6900k" is for sure good for AMD's marketing. I am feeling more optimistic for sure, but as with any new release I'm gonna wait till retail availability before making any conclusions.
 
I wonder where everyone else was. Looks Lisa put her face on this presentation. I am sure prior marketing failures led her to be like just let me do this shit. It did look like she was pretty damn confident in IPC numbers when she was mentioning it so I am sure they over delivered here. They can use some of that after under delivering with polaris, let me rephrase that. After fucking up polaris so bad. lol
 
CzlgRTFWEAIBcbJ.jpg


Well... Someone has some explaining to do.

My 5820K at 4350 did 52s, a 5960x at 4125mhz did 44s so how the hell is a stock 6900k at 29s ???
.eJwFwVEOgyAMANC7cAAKxVbmbQgSNFNKoGYfy-6-977mGZfZzKHa5wawnzPL2O1UGakWW0XqVVI_p81yQ1JN-bhL0wk-Ei8rEjmMjC9yBEgxMAaPC4e4sncMT3s3-TTbWzW_PwewIt0.Na_beLajqw6y5xPXS-vv1u7Ml_Y

Yep it doesn't add up. Plus the load is very light. The time is 36 seconds (From Kyle)

I got 77 seconds on a 4Ghz Skylake. Over twice the time.
 
Absolutely retarded name. Promising performance, but we won't know for sure until reviewers test them in the wild. AMD's marketing can't be trusted to present impartial data, and we don't know what special controls, or custom builds may have been used. They have a long history of pulling this shit at press releases, so time will tell.

I'm still concerned about the clocks. Yes, they are comparing it to Broadwell-E with similar core count and clocks, but we all know that's really not what matters. More cores don't make up for per core performence in 99.99 percent of workloads.

To me it depends on what kind of clocks it was actually running at during the broadwell comparison, and what the max we will hit in real overclocking is. Overclocks to the high 4's ghz is going to be absolutely crucial.
 
Absolutely retarded name. Promising performance, but we won't know for sure until reviewers test them in the wild. AMD's marketing can't be trusted to present impartial data, and we don't know what special controls, or custom builds may have been used. They have a long history of pulling this shit at press releases, so time will tell.

I'm still concerned about the clocks. Yes, they are comparing it to Broadwell-E with similar core count and clocks, but we all know that's really not what matters. More cores don't make up for per core performence in 99.99 percent of workloads.

To me it depends on what kind of clocks it was actually running at during the broadwell comparison, and what the max we will hit in real overclocking is. Overclocks to the high 4's ghz is going to be absolutely crucial.

It needs to be tested, that's for sure. Curious about this neural network branch prediction or whatever, sounds like marketing BS but I'm willing to be surprised
 
Absolutely retarded name. Promising performance, but we won't know for sure until reviewers test them in the wild. AMD's marketing can't be trusted to present impartial data, and we don't know what special controls, or custom builds may have been used. They have a long history of pulling this shit at press releases, so time will tell.

I'm still concerned about the clocks. Yes, they are comparing it to Broadwell-E with similar core count and clocks, but we all know that's really not what matters. More cores don't make up for per core performence in 99.99 percent of workloads.

To me it depends on what kind of clocks it was actually running at during the broadwell comparison, and what the max we will hit in real overclocking is. Overclocks to the high 4's ghz is going to be absolutely crucial.
So I am not convinced high 4ghz clocks matter that much if single thread performance is solid. Multi threaded apps dont scale linearly with core clocks
 
Machine learning just means more advanced--self training--statistical methods for brand prediction. It's the buzzword way of saying better cache hit/miss ratio.

Machine learning is machine learning, more accurate branch prediction and caching algorithms have nothing to do with it >_> technical marketing gobbledygook is a scourge
 
Back
Top