Ryzen 7 2700 seriously slow

Page 10 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

inf64

Diamond Member
Mar 11, 2011
3,685
3,957
136
If I was him, at this point I would just sell the system and go CF i5 or something similar. Prices will drop due to competitive pressure from Zen2 so CF i5 will become cheaper soon.
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
If I was him, at this point I would just sell the system and go CF i5 or something similar. Prices will drop due to competitive pressure from Zen2 so CF i5 will become cheaper soon.
Intel doesn't lower prices so a cheaper CF i5 is a very slim possibility.
They could release a new spin in a few months with more cores for the same price though,that's something they already did.
 

Wuzup101

Platinum Member
Feb 20, 2002
2,334
37
91
He already said it multiple times,he doesn't want to overclock it to get the performance he was suggested that he would get.
You even quoted it.

It was never advertised / reviewed as being 200-300mhz slower than the 2700X under heavy loads. It does sound like something is wrong with his setup, as he should get 4.1ghz with light single core loads, but the 2700 was shown to drop off speed wise (more towards 3400mhz), when more than 2 cores were loaded heavily, where the 2700X maintains much higher clocks stock under heavy workloads. The 2700X does drop off as more cores are loaded, but it's a gradual linear drop off, where as the 2700 drops off sharply after 2 cores are fully loaded.

Like I said, I have the same setup. I ran it stock for all of about an hour, and it did behave as expected. It runs around 3400mhz under heavy loads if I don't touch anything (heavy loads like 16 threads of prime 95).
 

killster1

Banned
Mar 15, 2007
6,208
475
126
Well, the timings are really loose, like 18-22-22-40 or something. But you're right, it could be the RAM. I tested it for a week at the default XMP settings that the ASUS UEFI set (3200Mhz), and it seemed stable, then I bumped it at the last moment to 3400, before I put the PC where it is now. It's doing DC work, and if the RAM was really marginal, though, I would expect to see errors or appcrashes or something.
U ran memtest for a week straight?
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
Put together a build using the much-maligned ASRock X470 Master SLI/ac and Ryzen 7 2700. Out of the box, it boosts to 4.1GHz on ST workloads, and 3.4GHz + up to 200MHz extra on MT workloads of short duration. After saturating the stock heatsink with a full 16-thread workload for an hour+, it stays at a steady 3400MHz.

So it appears to be working as reported elsewhere despite using stock cooler on a board with mediocre VRMs.
 

killster1

Banned
Mar 15, 2007
6,208
475
126
Did I specifically say I ran "Memtest" for a week straight? No. I "tested" it for a week straight. Running some DC apps and some web browsing and whatnot. Some benchmarks too.

well how long did you run memtest for. isnt that the only way to test ram? play games on it for a week and no bsod! or "i ran memtest for 48 hours straight and 0 errors"
 

VirtualLarry

No Lifer
Aug 25, 2001
56,227
9,990
126
Or, "Run a memory/CPU-intensive DC app on nearly all cores/threads for a week straight, and throw in some browsing too, and see if it BSODs or crashes".
 

killster1

Banned
Mar 15, 2007
6,208
475
126
Or, "Run a memory/CPU-intensive DC app on nearly all cores/threads for a week straight, and throw in some browsing too, and see if it BSODs or crashes".

since when has any review site ran "dc app" for a week to check their overclock stability? heh i dont even know what a DC app is ;( but i know i run memtest on everything, i love knowing stability is not from the ram.
 

killster1

Banned
Mar 15, 2007
6,208
475
126
Who said I was a "review site"?

i guess this is going no where. my point is review sites do it because it proves stability. if you want to fake stability then mention testing on apps that are not memtest. if you want to test true stability then u will test with memtest, end of discussion unless you can show me another program that will give definitive results!
 

VirtualLarry

No Lifer
Aug 25, 2001
56,227
9,990
126
if you want to fake stability then mention testing on apps that are not memtest. if you want to test true stability then u will test with memtest
Err, you realize, Memtest is NOT a CPU tester. I wanted to test the "whole package" (CPU OC and high RAM clock). Running DC for a week is a pretty good way to stress-test it, IMHO.

If you want to stick with Memtest, be my guest, I'm not disagreeing with you. Just sharing my methods. Good enough for me, at least. Surely, testing an OC, should involve stress-testing, as well as testing your desired app. Since DC work generally operates with a Quorum of 2 or more, I can go back and check my results and see if they have been spitting out anything faulty. Just like LinX, just over a longer period of time.

Edit: You DO realize, "Prime95", that got adopted and used by the OC'ing community as a "Stress Tester", is in fact, a "DC App", which is its primary purpose, not stress-testing.
 

killster1

Banned
Mar 15, 2007
6,208
475
126
Err, you realize, Memtest is NOT a CPU tester. I wanted to test the "whole package" (CPU OC and high RAM clock). Running DC for a week is a pretty good way to stress-test it, IMHO.

If you want to stick with Memtest, be my guest, I'm not disagreeing with you. Just sharing my methods. Good enough for me, at least. Surely, testing an OC, should involve stress-testing, as well as testing your desired app. Since DC work generally operates with a Quorum of 2 or more, I can go back and check my results and see if they have been spitting out anything faulty. Just like LinX, just over a longer period of time.

Edit: You DO realize, "Prime95", that got adopted and used by the OC'ing community as a "Stress Tester", is in fact, a "DC App", which is its primary purpose, not stress-testing.

if your memory is being used by a operating system it can not be tested right? whats your hate for memtest? who said cpu it was purely a ram conversation. If you dont care then i dont care too but id trust nothing less then 24 hours str8 of memtest
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,482
14,434
136
if your memory is being used by a operating system it can not be tested right? whats your hate for memtest? who said cpu it was purely a ram conversation. If you dont care then i dont care too but id trust nothing less then 24 hours str8 of memtest
memtest does fine for memory. It does not test the cpu or other parts of the system. Running "DC" (distributed computing ) applications tests all portions of a system most of the time. It errors out if there is a problem. memtest is only for determining if memory IS the problem.
 
  • Like
Reactions: VirtualLarry

killster1

Banned
Mar 15, 2007
6,208
475
126
memtest does fine for memory. It does not test the cpu or other parts of the system. Running "DC" (distributed computing ) applications tests all portions of a system most of the time. It errors out if there is a problem. memtest is only for determining if memory IS the problem.
"most of the time" how can you test memory you are using already? Never heard of anyone testing ram by running mining for bitcoin or what ever but i guess thats what you do to test your ram stability? hehe
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,482
14,434
136
"most of the time" how can you test memory you are using already? Never heard of anyone testing ram by running mining for bitcoin or what ever but i guess thats what you do to test your ram stability? hehe
I said nothing about mining, but the point is there are several pieces of stability, RAM, CPU motherboard, disk, and RAM is only one of them. By running DC software that checks the results, (BOINC), it tells you if the entire platform is working.
 

killster1

Banned
Mar 15, 2007
6,208
475
126
I said nothing about mining, but the point is there are several pieces of stability, RAM, CPU motherboard, disk, and RAM is only one of them. By running DC software that checks the results, (BOINC), it tells you if the entire platform is working.

If you say so, but the one thing i keep asking is if you can test memory that you are using already (the whole point of memtest) sure you can use other methods for 98% accuracy? i must repeat myself this was a RAM only test nothing else so not sure why you keep mentioning cpu motherboard disk..
 
Last edited:

VirtualLarry

No Lifer
Aug 25, 2001
56,227
9,990
126
i must repeat myself this was a RAM only test nothing else
I never said that, so I'm not sure where you're getting that from. You have some reading comprehension issues.

Well, the timings are really loose, like 18-22-22-40 or something. But you're right, it could be the RAM. I tested it (*) for a week at the default XMP settings that the ASUS UEFI set (3200Mhz), and it seemed stable, then I bumped it at the last moment to 3400, before I put the PC where it is now. It's doing DC work, and if the RAM was really marginal, though, I would expect to see errors or appcrashes or something.
(*) "It" being the entire rig, NOT just the RAM. Note later when I refer to the same "It" - "It's doing DC work". Showing that I meant that "It" refers to the entire system.

Edit: And "doing DC work all week" is equivalent to running Prime95 all week, which some people do to test RAM and CPU. Which is effectively what I was doing. (But using a different DC app than Prime95.)
 
Last edited:

killster1

Banned
Mar 15, 2007
6,208
475
126
I never said that, so I'm not sure where you're getting that from. You have some reading comprehension issues.

that was the whole conversation, me asking you about your "testing of the ram clock setting" amazing that you forgot.. its ok no big deal really this has gone to far i feel like i got trolled but its ok too still a big smile, was truely hoping for a better test than memtest to be mentioned, So bitcoin mining isnt a DC app? I havnt tested my system for bitcoin stability maybe anandtech should add that to the review table!!

"Stable overclock mined @ 2 weeks 0 bsod!" of course only 4gb of the 64gb ram was utilized for this test so..
 

VirtualLarry

No Lifer
Aug 25, 2001
56,227
9,990
126
i feel like i got trolled
LOL. I was sharing my knowledge, about my OC's 2700 rig, and my testing procedures, and you were picking at me and trying to prove that I was doing something wrong, for an entire forum page. :(

heh i dont even know what a DC app is ;(
No, just sadly ignorant of what we were discussing.

PS. I never mentioned testing with mining or "bitcoin mining". So you were throwing Straw Men at me likewise. So who's the real troll here?
 
Last edited:

killster1

Banned
Mar 15, 2007
6,208
475
126
LOL. I was sharing my knowledge, about my OC's 2700 rig, and my testing procedures, and you were picking at me and trying to prove that I was doing something wrong, for an entire forum page. :(


No, just sadly ignorant of what we were discussing.

PS. I never mentioned testing with mining or "bitcoin mining". So you were throwing Straw Men at me likewise. So who's the real troll here?



is bitcoin mining a dc app?
can you test your ram 100% when you are using it?
how much ram is used when bitcoin mining or what ever you are doing? 100%? doubtful / impossible?

Quit derailing this thread. You're on vacation for a week.

AT Moderator ElFenix
 
Last edited by a moderator:

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
is bitcoin mining a dc app?
can you test your ram 100% when you are using it?
how much ram is used when bitcoin mining or what ever you are doing? 100%? doubtful / impossible?
Okay you seem to have confusion on why these tests are run and how things work. If you want to test a stick of memory to find out if it has a bad sector (or more) you use Memtest or OEM specific memory testers. It will go through and test a few types of comminication styles to the memory to see if it can trip up. But you do that stock settings. You can run them overclocked but these tests are mainly to each specific block of space the memory has for informaition.

Prime 95 and its ilk (which can include certain types of Bitcoin minning) are system testers. They don't stress video outside Bitcoin mining (certain forms). But what they do is have the CPU running at 100% constantly going back and forth to memory. This puts heavy strain on the CPU and on top of that the IMC and the Memory. So that if there is a bad actor in the chain the system locks up and everybody is angry. Then you use specific tools to figure out who is acting out. This doesn't just stress the actual memory sectors themselves (but bad sectors can trip the system up) but will test the whole of the memory access sub system. Which is where faults usually lie when overclocking after you have tested the memory originally. Usually you know if you have a decent stable system after 24 hrs of this kind of burnin. Guys who do it for a week are just masochists.
 
  • Like
Reactions: VirtualLarry

fkoehler

Member
Feb 29, 2008
193
145
116
This thread is still going?
Haven't read all responses yet, however has the OP yet done any simple substitution troubleshooting to start ruling out sub-systems?
If not, then it seems like the purpose is just to rag on AMD for some reason...
 
  • Like
Reactions: MangoX

killster1

Banned
Mar 15, 2007
6,208
475
126
Okay you seem to have confusion on why these tests are run and how things work. If you want to test a stick of memory to find out if it has a bad sector (or more) you use Memtest or OEM specific memory testers. It will go through and test a few types of comminication styles to the memory to see if it can trip up. But you do that stock settings. You can run them overclocked but these tests are mainly to each specific block of space the memory has for informaition.

Prime 95 and its ilk (which can include certain types of Bitcoin minning) are system testers. They don't stress video outside Bitcoin mining (certain forms). But what they do is have the CPU running at 100% constantly going back and forth to memory. This puts heavy strain on the CPU and on top of that the IMC and the Memory. So that if there is a bad actor in the chain the system locks up and everybody is angry. Then you use specific tools to figure out who is acting out. This doesn't just stress the actual memory sectors themselves (but bad sectors can trip the system up) but will test the whole of the memory access sub system. Which is where faults usually lie when overclocking after you have tested the memory originally. Usually you know if you have a decent stable system after 24 hrs of this kind of burnin. Guys who do it for a week are just masochists.


how much ram is used? why not test all the ram if talking about stable overclock for ram. i guess i can just keep asking over and over how do you test memory that is in use by operating system already...